How AI models score on clinical work

• HealthBench Professional0.51851.8×0.743length-adjusted, max reasoning effort, GPT-5.6 system card Table 6 column GPT-5.5 (57.2 unadjusted, 3818 chars)
• MedScribe (Vals AI)86.8786.9×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25
• MedCode (Vals AI)49.1049.1×0.911model ID openai/gpt-5.5; reasoning_effort=xhigh; temperature=1; max_output_tokens=30000
• First, Do NOHARM (v2)70.070.0×0.507
• PhysicianBench46.346.3×0.79pass@1; Pass^3 28.0

Weighted average of 5 boards

55.3

Multiplier for 5 of 10 boards

× 0.913

Index

55.3 × 0.913 = 50.5

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• HealthBench Professional0.62162.1×0.743length-adjusted (method published in the HealthBench Professional paper); Anthropic protocol: adaptive thinking at max effort, Claude Opus 4.8 grader, averaged over five trials, no tools or customized system prompt; Fable 5.1 run with safety classifiers active and a refusal-fallback to Claude Opus 5 (raw 74.2%).
• MedScribe (Vals AI)91.2991.3×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25
• MedCode (Vals AI)53.5153.5×0.911
• PhysicianBench61.061.0×0.79Anthropic harness; 100 tasks; pass@1; max effort; Opus 5 rubric grader; safety classifiers enabled

Weighted average of 4 boards

61.6

Multiplier for 4 of 10 boards

× 0.816

Index

61.6 × 0.816 = 50.3

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• HealthBench Professional0.57457.4×0.743Anthropic June evaluation; length-adjusted; adaptive max effort; Opus 4.8 grader; five trials; no tools or custom system prompt
• MedScribe (Vals AI)85.7585.8×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID anthropic/claude-opus-4-8; temperature=1; max_output_tokens=30000; compute_effort=max
• MedCode (Vals AI)53.2253.2×0.911
• MedXpertQA (MM)71.771.7×0.463, near its ceilingNear its ceiling: leader 81.5, weight 0.463Qwen-run comparison in the Qwen3.8-Max launch post

Weighted average of 4 boards

61.6

Multiplier for 4 of 10 boards

× 0.816

Index

61.6 × 0.816 = 50.3

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedXpertQA (MM)70.070.0×0.463, near its ceilingNear its ceiling: leader 81.5, weight 0.463self-reported in the Qwen3.5-397B-A17B model card
• MAST (Medical AI Superintelligence Test)57.957.9×0.995
• First, Do NOHARM (v2)61.161.1×0.507
• EHR-Complex0.62062.0×0.875headline evaluation

Weighted average of 4 boards

61.7

Multiplier for 4 of 10 boards

× 0.816

Index

61.7 × 0.816 = 50.3

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• HealthBench Professional0.59359.3×0.743length-normalized, GPT-5.4 low-reasoning grader, xhigh reasoning via Meta Model API (Muse Spark 1.1 Evaluation Report Figure 44)
• MedScribe (Vals AI)88.8988.9×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25
• First, Do NOHARM (v2)79.779.7×0.507

Weighted average of 3 boards

71.1

Multiplier for 3 of 10 boards

× 0.707

Index

71.1 × 0.707 = 50.3

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)87.9688.0×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25
• MedCode (Vals AI)48.8848.9×0.911model ID kimi/kimi-k3; temperature=1; max_output_tokens=30000
• MAST (Medical AI Superintelligence Test)60.160.1×0.995
• First, Do NOHARM (v2)74.074.0×0.507

Weighted average of 4 boards

61.5

Multiplier for 4 of 10 boards

× 0.816

Index

61.5 × 0.816 = 50.2

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)73.5573.6×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID google/gemini-2.5-pro; temperature=1; max_output_tokens=30000
• MedCode (Vals AI)50.5950.6×0.911model ID google/gemini-2.5-pro; temperature=1; max_output_tokens=30000
• MedPIC54.454.4×0.482, near its ceilingNear its ceiling: leader 80.7, weight 0.482zero-shot; independent questions; exact option-set match
• MedHELM0.52952.9×0.87
• First, Do NOHARM (v2)61.961.9×0.507
• EHR-Complex0.31031.0×0.875Table 3; macro-average across 12 intent/scope columns; temperature 0; up to 50 SQL/Python interaction turns

Weighted average of 6 boards

50.1

Multiplier for 6 of 10 boards

× 1

Index

50.1 × 1 = 50.1

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• HealthBench Professional0.57857.8×0.743length-adjusted, Anthropic protocol: adaptive thinking at max effort, Claude Opus 4.8 grader, averaged over 5 trials, no tools or custom system prompt (raw 62.4%).
• MedScribe (Vals AI)76.0576.1×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID anthropic/claude-sonnet-5; temperature=1; max_output_tokens=30000; compute_effort=max
• MedCode (Vals AI)47.5447.5×0.911model ID anthropic/claude-sonnet-5; compute_effort=max; temperature=1; max_output_tokens=30000
• MAST (Medical AI Superintelligence Test)56.656.6×0.995
• PhysicianBench37.437.4×0.79Anthropic harness; 100 tasks; pass@1; max effort; Opus 5 rubric grader; safety classifiers enabled

Weighted average of 5 boards

51.8

Multiplier for 5 of 10 boards

× 0.913

Index

51.8 × 0.913 = 47.3

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)76.4476.4×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID kimi/kimi-k2.5-thinking; top_p=0.95; max_output_tokens=30000
• MedCode (Vals AI)39.3239.3×0.911model ID kimi/kimi-k2.5-thinking; top_p=0.95; max_output_tokens=30000
• MedXpertQA (MM)65.365.3×0.463, near its ceilingNear its ceiling: leader 81.5, weight 0.463Qwen-run comparison in the Qwen3.5-397B-A17B model card (K2.5-1T-A32B column)
• EHR-Complex0.62062.0×0.875headline 12-model evaluation, top open-weight

Weighted average of 4 boards

55.8

Multiplier for 4 of 10 boards

× 0.816

Index

55.8 × 0.816 = 45.5

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)86.7486.7×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25
• MedCode (Vals AI)49.1349.1×0.911model ID anthropic/claude-opus-4-6-thinking; compute_effort=max; temperature=1; max_output_tokens=30000
• MedXpertQA (MM)64.864.8×0.463, near its ceilingNear its ceiling: leader 81.5, weight 0.463Meta's Muse Spark launch table (Meta reports the better of vendor self-reports and its own reproduction)
• MedHELM0.45645.6×0.87
• PhysicianBench31.731.7×0.79Pass^3 18.0

Weighted average of 5 boards

49.1

Multiplier for 5 of 10 boards

× 0.913

Index

49.1 × 0.913 = 44.8

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• HealthBench Professional0.70370.3×0.743Anthropic public-API reproduction; max effort; no system prompt; Opus 4.8 grader; length-adjusted; raw 74.0%
• MedScribe (Vals AI)87.9187.9×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25
• MedCode (Vals AI)48.4948.5×0.911

Weighted average of 3 boards

62.2

Multiplier for 3 of 10 boards

× 0.707

Index

62.2 × 0.707 = 44.0

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)76.5776.6×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID google/gemini-3.5-flash; temperature=1; max_output_tokens=30000; reasoning_effort=high
• MedCode (Vals AI)55.8355.8×0.911
• MedHELM0.64264.2×0.87

Weighted average of 3 boards

62.0

Multiplier for 3 of 10 boards

× 0.707

Index

62.0 × 0.707 = 43.8

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)72.0472.0×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID google/gemini-3-pro-preview; temperature=1; max_output_tokens=30000; reasoning_effort=high
• MedXpertQA (MM)76.076.0×0.463, near its ceilingNear its ceiling: leader 81.5, weight 0.463Qwen3.5 model card comparison; source-specific evaluation, not harmonized across vendors

Weighted average of 2 boards

74.6

Multiplier for 2 of 10 boards

× 0.577

Index

74.6 × 0.577 = 43.0

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• HealthBench Professional0.64264.2×0.743OpenAI; length-adjusted; 64.2 (67.2, 3038) (adjusted, raw, mean response characters); addendum Table 7 refers to the GPT-6 Astra card for method
• MedScribe (Vals AI)86.4586.5×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID openai/gpt-6.1-sol; max_output_tokens=128000; reasoning_effort=max
• MedCode (Vals AI)48.8448.8×0.911model ID openai/gpt-6.1-sol; reasoning_effort=max; max_output_tokens=128000

Weighted average of 3 boards

59.8

Multiplier for 3 of 10 boards

× 0.707

Index

59.8 × 0.707 = 42.3

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)79.6679.7×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID google/gemini-3.6-flash; temperature=1; max_output_tokens=30000; reasoning_effort=high
• MedCode (Vals AI)53.1553.2×0.911
• MAST (Medical AI Superintelligence Test)59.359.3×0.995

Weighted average of 3 boards

59.1

Multiplier for 3 of 10 boards

× 0.707

Index

59.1 × 0.707 = 41.8

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)85.3285.3×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID anthropic/claude-opus-4-5-20251101-thinking; temperature=1; max_output_tokens=30000; compute_effort=high
• MedCode (Vals AI)49.1649.2×0.911model ID anthropic/claude-opus-4-5-20251101-thinking; compute_effort=high; temperature=1; max_output_tokens=30000
• MedXpertQA (MM)63.663.6×0.463, near its ceilingNear its ceiling: leader 81.5, weight 0.463Qwen3.5 model card comparison; source-specific evaluation, not harmonized across vendors

Weighted average of 3 boards

58.9

Multiplier for 3 of 10 boards

× 0.707

Index

58.9 × 0.707 = 41.6

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)84.9585.0×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID alibaba/qwen3.8-max; temperature=1; max_output_tokens=30000
• MedCode (Vals AI)40.6740.7×0.911model ID alibaba/qwen3.8-max; temperature=1; max_output_tokens=30000
• MedXpertQA (MM)80.480.4×0.463, near its ceilingNear its ceiling: leader 81.5, weight 0.463Alibaba's own Qwen3.8 launch table; protocol not stated

Weighted average of 3 boards

58.8

Multiplier for 3 of 10 boards

× 0.707

Index

58.8 × 0.707 = 41.6

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• HealthBench Professional0.56756.7×0.743SpaceXAI evaluation; xHigh effort; HealthBench Professional
• MedScribe (Vals AI)89.3889.4×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID grok/grok-4.7; temperature=1; top_p=0.95; reasoning_effort=xhigh
• MedCode (Vals AI)49.5549.6×0.911model ID grok/grok-4.7; reasoning_effort=xhigh; temperature=1; top_p=0.95

Weighted average of 3 boards

57.6

Multiplier for 3 of 10 boards

× 0.707

Index

57.6 × 0.707 = 40.7

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• HealthBench Professional0.60860.8×0.743OpenAI; length-adjusted; maximum reasoning effort; 60.8 (59.5, 1573) (adjusted, raw, mean response characters)
• MedScribe (Vals AI)82.0382.0×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID openai/gpt-6-sol; max_output_tokens=128000; reasoning_effort=max
• MedCode (Vals AI)47.0747.1×0.911model ID openai/gpt-6-sol; reasoning_effort=max; max_output_tokens=128000

Weighted average of 3 boards

57.0

Multiplier for 3 of 10 boards

× 0.707

Index

57.0 × 0.707 = 40.3

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• HealthBench Professional0.51951.9×0.743length-adjusted; adaptive thinking at max effort; Claude Sonnet 4.6 grader; 5 trials; comparison model in the Opus 4.8 card
• MedScribe (Vals AI)82.9583.0×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID anthropic/claude-opus-4-7; temperature=1; max_output_tokens=30000; compute_effort=max
• MedCode (Vals AI)54.8654.9×0.911
• PhysicianBench29.329.3×0.79Pass^3 18.0

Weighted average of 4 boards

49.2

Multiplier for 4 of 10 boards

× 0.816

Index

49.2 × 0.816 = 40.1

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• HealthBench Professional0.60860.8×0.743OpenAI; length-adjusted; maximum reasoning effort; 60.8 (61.2, 2119) (adjusted, raw, mean response characters)
• MedScribe (Vals AI)83.7183.7×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID openai/gpt-6-luna; max_output_tokens=128000; reasoning_effort=max
• MedCode (Vals AI)44.6944.7×0.911model ID openai/gpt-6-luna; reasoning_effort=max; max_output_tokens=128000

Weighted average of 3 boards

56.1

Multiplier for 3 of 10 boards

× 0.707

Index

56.1 × 0.707 = 39.7

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• HealthBench Professional0.57757.7×0.743length-adjusted, max reasoning effort (62.4 unadjusted, 3,618 mean response chars); GPT-5.6 system card Table 6, column GPT-5.6-TERRA.
• MedScribe (Vals AI)82.8782.9×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID openai/gpt-5.6-terra; max_output_tokens=30000; reasoning_effort=xhigh
• MedCode (Vals AI)43.4143.4×0.911model ID openai/gpt-5.6-terra; reasoning_effort=xhigh; max_output_tokens=30000

Weighted average of 3 boards

54.2

Multiplier for 3 of 10 boards

× 0.707

Index

54.2 × 0.707 = 38.3

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• HealthBench Professional0.55755.7×0.743length-adjusted, max reasoning effort (59.8 unadjusted, 3,389 mean response chars); GPT-5.6 system card Table 6, column GPT-5.6-LUNA. API max-effort setting, not the August ChatGPT production setting measured in row 493.
• MedScribe (Vals AI)84.3984.4×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID openai/gpt-5.6-luna; max_output_tokens=30000; reasoning_effort=max
• MedCode (Vals AI)42.3942.4×0.911model ID openai/gpt-5.6-luna; reasoning_effort=max; max_output_tokens=30000

Weighted average of 3 boards

53.1

Multiplier for 3 of 10 boards

× 0.707

Index

53.1 × 0.707 = 37.5

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• HealthBench Professional0.39639.6×0.743length-adjusted, max reasoning effort, GPT-5.6 system card Table 6 column GPT-5.1 (48.0 unadjusted, 4863 chars)
• MedScribe (Vals AI)88.0988.1×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25
• MedCode (Vals AI)52.7352.7×0.911

Weighted average of 3 boards

52.2

Multiplier for 3 of 10 boards

× 0.707

Index

52.2 × 0.707 = 36.9

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• HealthBench Professional0.48548.5×0.743SpaceXAI evaluation; High effort; HealthBench Professional
• MedScribe (Vals AI)86.5386.5×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25
• MedCode (Vals AI)44.7144.7×0.911model ID grok/grok-4.6; reasoning_effort=high; temperature=1; top_p=0.95; max_output_tokens=30000

Weighted average of 3 boards

51.7

Multiplier for 3 of 10 boards

× 0.707

Index

51.7 × 0.707 = 36.6

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)88.9488.9×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID zai/glm-5.3-flash; temperature=1; top_p=0.95; max_output_tokens=30000; reasoning_effort=max

One board, so its score

88.9

Multiplier for 1 of 10 boards

× 0.408

Index

88.9 × 0.408 = 36.3

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)80.1780.2×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID deepseek/deepseek-v4-pro-0813; max_output_tokens=30000; reasoning_effort=max
• MedCode (Vals AI)42.4742.5×0.911model ID deepseek/deepseek-v4-pro-0813; reasoning_effort=max; max_output_tokens=30000
• MedPIC71.771.7×0.482, near its ceilingNear its ceiling: leader 80.7, weight 0.482zero-shot; independent questions; exact option-set match
• PhysicianBench18.718.7×0.79Pass@1 over 3 runs; Pass^3 6.0; shared FHIR tool harness; up to 100 turns; high reasoning when supported

Weighted average of 4 boards

44.4

Multiplier for 4 of 10 boards

× 0.816

Index

44.4 × 0.816 = 36.2

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)72.2772.3×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID zai/glm-5.1; temperature=1; top_p=0.95; max_output_tokens=30000
• MedCode (Vals AI)41.6041.6×0.911model ID zai/glm-5.1; temperature=1; top_p=0.95; max_output_tokens=30000
• First, Do NOHARM (v2)57.957.9×0.507First Do NOHARM v2 overall; preview; open-weight model; run date unpublished

Weighted average of 3 boards

51.2

Multiplier for 3 of 10 boards

× 0.707

Index

51.2 × 0.707 = 36.2

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)74.4074.4×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID grok/grok-4.3; temperature=1; top_p=0.95; max_output_tokens=30000
• MedCode (Vals AI)38.0738.1×0.911model ID grok/grok-4.3; temperature=1; top_p=0.95; max_output_tokens=30000
• MAST (Medical AI Superintelligence Test)53.753.7×0.995

Weighted average of 3 boards

49.5

Multiplier for 3 of 10 boards

× 0.707

Index

49.5 × 0.707 = 35.0

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)83.9483.9×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID google/gemini-3.7-flash; temperature=1; max_output_tokens=30000; reasoning_effort=high
• MedCode (Vals AI)53.3953.4×0.911model ID google/gemini-3.7-flash; reasoning_effort=high; temperature=1; max_output_tokens=30000

Weighted average of 2 boards

60.0

Multiplier for 2 of 10 boards

× 0.577

Index

60.0 × 0.577 = 34.6

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)69.9269.9×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID google/gemini-3-flash-preview; temperature=1; max_output_tokens=30000; reasoning_effort=high
• MedCode (Vals AI)55.9255.9×0.911

Weighted average of 2 boards

58.9

Multiplier for 2 of 10 boards

× 0.577

Index

58.9 × 0.577 = 34.0

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)82.9883.0×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID google/gemini-2.5-flash-thinking; temperature=1; max_output_tokens=30000

One board, so its score

83.0

Multiplier for 1 of 10 boards

× 0.408

Index

83.0 × 0.408 = 33.9

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)90.0690.1×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25
• MedCode (Vals AI)49.3549.4×0.911model ID meta/muse_spark_1_2; reasoning_effort=xhigh; temperature=1; max_output_tokens=30000

Weighted average of 2 boards

58.2

Multiplier for 2 of 10 boards

× 0.577

Index

58.2 × 0.577 = 33.6

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)78.1578.2×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID kimi/kimi-k2.6; top_p=0.95; max_output_tokens=30000
• MedCode (Vals AI)40.1440.1×0.911model ID kimi/kimi-k2.6; top_p=0.95; max_output_tokens=30000
• First, Do NOHARM (v2)59.159.1×0.507
• PhysicianBench17.017.0×0.79open source

Weighted average of 4 boards

40.5

Multiplier for 4 of 10 boards

× 0.816

Index

40.5 × 0.816 = 33.0

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)80.3680.4×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID deepseek/deepseek-v4-flash-0731; max_output_tokens=30000; reasoning_effort=high

One board, so its score

80.4

Multiplier for 1 of 10 boards

× 0.408

Index

80.4 × 0.408 = 32.8

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)76.9677.0×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID alibaba/qwen3.6-plus; temperature=1; max_output_tokens=30000
• MedCode (Vals AI)36.8936.9×0.911model ID alibaba/qwen3.6-plus; temperature=1; max_output_tokens=30000
• MedXpertQA (MM)68.768.7×0.463, near its ceilingNear its ceiling: leader 81.5, weight 0.463Qwen-run comparison in the Qwen3.7-Plus launch post
• PhysicianBench13.713.7×0.79

Weighted average of 4 boards

39.6

Multiplier for 4 of 10 boards

× 0.816

Index

39.6 × 0.816 = 32.3

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)84.5084.5×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID google/gemini-3.8-flash; temperature=1; max_output_tokens=65536; reasoning_effort=high
• MedCode (Vals AI)48.1348.1×0.911model ID google/gemini-3.8-flash; reasoning_effort=high; temperature=1; max_output_tokens=65536

Weighted average of 2 boards

55.9

Multiplier for 2 of 10 boards

× 0.577

Index

55.9 × 0.577 = 32.3

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• HealthBench Professional0.44244.2×0.743length-adjusted; Anthropic protocol: adaptive thinking at max effort, Claude Opus 4.8 grader, averaged over 5 trials, no tools or custom system prompt; comparison column in the Sonnet 5 card. Other Anthropic prints: 44.4% (Fable card Figure 8.18.2.A), 41.7% (Opus 4.8 card, Sonnet 4.6 grader)
• MedPIC64.264.2×0.482, near its ceilingNear its ceiling: leader 80.7, weight 0.482zero-shot; independent questions; exact option-set match
• PhysicianBench23.023.0×0.79
• EHR-Complex0.36036.0×0.875validation configuration

Weighted average of 4 boards

39.3

Multiplier for 4 of 10 boards

× 0.816

Index

39.3 × 0.816 = 32.1

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)87.2587.3×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25
• MedCode (Vals AI)46.2946.3×0.911model ID minimax/MiniMax-M3; temperature=1; top_p=0.95; max_output_tokens=30000

Weighted average of 2 boards

55.1

Multiplier for 2 of 10 boards

× 0.577

Index

55.1 × 0.577 = 31.8

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)88.3188.3×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID xiaomi/mimo-v2.6-pro; temperature=1; top_p=0.95; max_output_tokens=128000
• MedCode (Vals AI)44.9745.0×0.911model ID xiaomi/mimo-v2.6-pro; temperature=1; top_p=0.95; max_output_tokens=128000

Weighted average of 2 boards

54.3

Multiplier for 2 of 10 boards

× 0.577

Index

54.3 × 0.577 = 31.3

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• HealthBench Professional0.64864.8×0.743Anthropic; max effort; Opus 4.8 grader; safety classifiers enabled; length-adjusted; raw 71.0%; five runs (October 1 and 2); thirteen repeat max runs on October 4 averaged 64.2%
• PhysicianBench43.043.0×0.79Anthropic harness; 100 tasks; pass@1; max effort; Opus 5 rubric grader; safety classifiers enabled; five attempts per task; run October 1 and 2

Weighted average of 2 boards

53.6

Multiplier for 2 of 10 boards

× 0.577

Index

53.6 × 0.577 = 30.9

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)76.6576.7×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID openai/o3-2025-04-16; max_output_tokens=30000; reasoning_effort=high
• MedCode (Vals AI)47.2947.3×0.911model ID openai/o3-2025-04-16; reasoning_effort=high; max_output_tokens=30000

Weighted average of 2 boards

53.6

Multiplier for 2 of 10 boards

× 0.577

Index

53.6 × 0.577 = 30.9

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)73.9073.9×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID anthropic/claude-opus-4-1-20250805-thinking; temperature=1; max_output_tokens=30000
• MedCode (Vals AI)47.2347.2×0.911model ID anthropic/claude-opus-4-1-20250805-thinking; temperature=1; max_output_tokens=30000

Weighted average of 2 boards

52.9

Multiplier for 2 of 10 boards

× 0.577

Index

52.9 × 0.577 = 30.5

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)84.5284.5×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID anthropic/claude-sonnet-4-5-20250929; temperature=1; max_output_tokens=30000
• MedCode (Vals AI)44.1344.1×0.911model ID anthropic/claude-sonnet-4-5-20250929-thinking; temperature=1; max_output_tokens=30000

Weighted average of 2 boards

52.8

Multiplier for 2 of 10 boards

× 0.577

Index

52.8 × 0.577 = 30.5

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)88.8188.8×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID zai/glm-5.3; temperature=1; top_p=0.95; max_output_tokens=30000; reasoning_effort=max
• MedCode (Vals AI)42.8642.9×0.911model ID zai/glm-5.3; reasoning_effort=max; temperature=1; top_p=0.95; max_output_tokens=30000

Weighted average of 2 boards

52.8

Multiplier for 2 of 10 boards

× 0.577

Index

52.8 × 0.577 = 30.5

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)86.8886.9×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID grok/grok-4.5; temperature=1; top_p=0.95; max_output_tokens=30000; reasoning_effort=high
• MedCode (Vals AI)43.2943.3×0.911model ID grok/grok-4.5; reasoning_effort=high; temperature=1; top_p=0.95; max_output_tokens=30000

Weighted average of 2 boards

52.7

Multiplier for 2 of 10 boards

× 0.577

Index

52.7 × 0.577 = 30.4

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)83.6083.6×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID tencent/hy4-preview; temperature=1; top_p=1; max_output_tokens=64000
• MedCode (Vals AI)43.2543.3×0.911model ID tencent/hy4-preview; temperature=1; top_p=1; max_output_tokens=64000

Weighted average of 2 boards

52.0

Multiplier for 2 of 10 boards

× 0.577

Index

52.0 × 0.577 = 30.0

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedHELM0.48548.5×0.87
• First, Do NOHARM (v2)55.855.8×0.507

Weighted average of 2 boards

51.2

Multiplier for 2 of 10 boards

× 0.577

Index

51.2 × 0.577 = 29.5

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)63.9063.9×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID google/gemini-3.1-flash-lite-preview; temperature=1; max_output_tokens=30000; reasoning_effort=high
• MedCode (Vals AI)47.6047.6×0.911model ID google/gemini-3.1-flash-lite-preview; reasoning_effort=high; temperature=1; max_output_tokens=30000

Weighted average of 2 boards

51.1

Multiplier for 2 of 10 boards

× 0.577

Index

51.1 × 0.577 = 29.5

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)80.5880.6×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID openai/gpt-5-mini-2025-08-07; max_output_tokens=30000; reasoning_effort=high
• MedCode (Vals AI)43.0543.1×0.911model ID openai/gpt-5-mini-2025-08-07; reasoning_effort=high; max_output_tokens=30000

Weighted average of 2 boards

51.2

Multiplier for 2 of 10 boards

× 0.577

Index

51.2 × 0.577 = 29.5

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedPIC72.472.4×0.482, near its ceilingNear its ceiling: leader 80.7, weight 0.482zero-shot; independent questions; exact option-set match

One board, so its score

72.4

Multiplier for 1 of 10 boards

× 0.408

Index

72.4 × 0.408 = 29.5

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)85.5085.5×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID deepseek/deepseek-v4.1-flash; max_output_tokens=384000; reasoning_effort=high
• MedCode (Vals AI)41.1741.2×0.911model ID deepseek/deepseek-v4.1-flash; reasoning_effort=high; max_output_tokens=384000

Weighted average of 2 boards

50.7

Multiplier for 2 of 10 boards

× 0.577

Index

50.7 × 0.577 = 29.3

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)85.4185.4×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID thinkingmachines/inkling; temperature=1; top_p=1; max_output_tokens=30000; reasoning_effort=0.99
• MedCode (Vals AI)41.1941.2×0.911model ID thinkingmachines/inkling; reasoning_effort=0.99; temperature=1; top_p=1; max_output_tokens=30000

Weighted average of 2 boards

50.7

Multiplier for 2 of 10 boards

× 0.577

Index

50.7 × 0.577 = 29.3

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)85.2885.3×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID xiaomi/mimo-v2.6-flash; temperature=1; top_p=0.95; max_output_tokens=128000
• MedCode (Vals AI)41.0641.1×0.911model ID xiaomi/mimo-v2.6-flash; temperature=1; top_p=0.95; max_output_tokens=128000

Weighted average of 2 boards

50.6

Multiplier for 2 of 10 boards

× 0.577

Index

50.6 × 0.577 = 29.2

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedXpertQA (MM)71.071.0×0.463, near its ceilingNear its ceiling: leader 81.5, weight 0.463Alibaba's own Qwen3.7 Plus launch table; protocol not stated

One board, so its score

71.0

Multiplier for 1 of 10 boards

× 0.408

Index

71.0 × 0.408 = 29.0

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)83.5383.5×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID zai/glm-5.2; temperature=1; max_output_tokens=30000
• MedCode (Vals AI)40.7740.8×0.911model ID zai/glm-5.2; temperature=1; max_output_tokens=30000

Weighted average of 2 boards

50.0

Multiplier for 2 of 10 boards

× 0.577

Index

50.0 × 0.577 = 28.9

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedPIC70.770.7×0.482, near its ceilingNear its ceiling: leader 80.7, weight 0.482zero-shot; independent questions; exact option-set match

One board, so its score

70.7

Multiplier for 1 of 10 boards

× 0.408

Index

70.7 × 0.408 = 28.8

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)70.8970.9×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID google/gemini-3.5-flash-lite; temperature=1; max_output_tokens=30000; reasoning_effort=high
• MedCode (Vals AI)43.4943.5×0.911model ID google/gemini-3.5-flash-lite; reasoning_effort=high; temperature=1; max_output_tokens=30000

Weighted average of 2 boards

49.4

Multiplier for 2 of 10 boards

× 0.577

Index

49.4 × 0.577 = 28.5

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedPIC69.469.4×0.482, near its ceilingNear its ceiling: leader 80.7, weight 0.482zero-shot; independent questions; exact option-set match

One board, so its score

69.4

Multiplier for 1 of 10 boards

× 0.408

Index

69.4 × 0.408 = 28.3

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)77.0977.1×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID openai/gpt-5.4-nano-2026-03-17; max_output_tokens=30000; reasoning_effort=high
• MedCode (Vals AI)41.0341.0×0.911model ID openai/gpt-5.4-nano-2026-03-17; reasoning_effort=high; max_output_tokens=30000

Weighted average of 2 boards

48.8

Multiplier for 2 of 10 boards

× 0.577

Index

48.8 × 0.577 = 28.2

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedPIC68.568.5×0.482, near its ceilingNear its ceiling: leader 80.7, weight 0.482zero-shot; independent questions; exact option-set match

One board, so its score

68.5

Multiplier for 1 of 10 boards

× 0.408

Index

68.5 × 0.408 = 27.9

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)84.1184.1×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID thinkingmachines/inkling-small; temperature=1; top_p=1; max_output_tokens=30000; reasoning_effort=0.99
• MedCode (Vals AI)37.8937.9×0.911model ID thinkingmachines/inkling-small; reasoning_effort=0.99; temperature=1; top_p=1; max_output_tokens=30000

Weighted average of 2 boards

47.8

Multiplier for 2 of 10 boards

× 0.577

Index

47.8 × 0.577 = 27.6

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)79.4079.4×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID alibaba/qwen3.7-max; temperature=1; max_output_tokens=30000
• MedCode (Vals AI)38.7538.8×0.911model ID alibaba/qwen3.7-max; temperature=1; max_output_tokens=30000

Weighted average of 2 boards

47.5

Multiplier for 2 of 10 boards

× 0.577

Index

47.5 × 0.577 = 27.4

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)81.6381.6×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID grok/grok-4-fast-reasoning; temperature=1; top_p=0.95; max_output_tokens=30000
• MedCode (Vals AI)37.3837.4×0.911model ID grok/grok-4-fast-reasoning; temperature=1; top_p=0.95; max_output_tokens=30000

Weighted average of 2 boards

46.9

Multiplier for 2 of 10 boards

× 0.577

Index

46.9 × 0.577 = 27.1

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)63.4163.4×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID grok/grok-4.20-0309-reasoning; temperature=1; top_p=0.95; max_output_tokens=30000
• MedCode (Vals AI)32.1632.2×0.911model ID grok/grok-4.20-0309-reasoning; temperature=1; top_p=0.95; max_output_tokens=30000
• MedXpertQA (MM)65.865.8×0.463, near its ceilingNear its ceiling: leader 81.5, weight 0.463Meta's Muse Spark launch table (Meta reports the better of vendor self-reports and its own reproduction)
• PhysicianBench5.35.3×0.79

Weighted average of 4 boards

33.1

Multiplier for 4 of 10 boards

× 0.816

Index

33.1 × 0.816 = 27.0

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)78.1578.2×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID grok/grok-4-0709; temperature=1; top_p=0.95; max_output_tokens=30000
• MedCode (Vals AI)38.0838.1×0.911model ID grok/grok-4-0709; temperature=1; top_p=0.95; max_output_tokens=30000

Weighted average of 2 boards

46.7

Multiplier for 2 of 10 boards

× 0.577

Index

46.7 × 0.577 = 26.9

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedPIC63.863.8×0.482, near its ceilingNear its ceiling: leader 80.7, weight 0.482zero-shot; independent questions; exact option-set match

One board, so its score

63.8

Multiplier for 1 of 10 boards

× 0.408

Index

63.8 × 0.408 = 26.0

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)80.7880.8×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID minimax/MiniMax-M2.1; temperature=1; top_p=0.95; max_output_tokens=30000
• MedCode (Vals AI)34.0834.1×0.911model ID minimax/MiniMax-M2.1; temperature=1; top_p=0.95; max_output_tokens=30000

Weighted average of 2 boards

44.2

Multiplier for 2 of 10 boards

× 0.577

Index

44.2 × 0.577 = 25.5

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedXpertQA (MM)61.361.3×0.463, near its ceilingNear its ceiling: leader 81.5, weight 0.463Google model card; MedXPertQA MM row; vendor-reported; protocol differs from other source families

One board, so its score

61.3

Multiplier for 1 of 10 boards

× 0.408

Index

61.3 × 0.408 = 25.0

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedPIC61.261.2×0.482, near its ceilingNear its ceiling: leader 80.7, weight 0.482zero-shot; independent questions; exact option-set match

One board, so its score

61.2

Multiplier for 1 of 10 boards

× 0.408

Index

61.2 × 0.408 = 25.0

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)72.4172.4×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID anthropic/claude-sonnet-4-20250514; temperature=1; max_output_tokens=30000
• MedCode (Vals AI)34.9635.0×0.911model ID anthropic/claude-sonnet-4-20250514-thinking; max_output_tokens=30000

Weighted average of 2 boards

43.1

Multiplier for 2 of 10 boards

× 0.577

Index

43.1 × 0.577 = 24.9

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)80.9080.9×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID ant/ling-3.0-flash-2607; temperature=1; top_p=0.95; max_output_tokens=30000
• MedCode (Vals AI)32.2732.3×0.911model ID ant/ling-3.0-flash-2607; temperature=1; top_p=0.95; max_output_tokens=30000

Weighted average of 2 boards

42.8

Multiplier for 2 of 10 boards

× 0.577

Index

42.8 × 0.577 = 24.7

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedPIC60.660.6×0.482, near its ceilingNear its ceiling: leader 80.7, weight 0.482zero-shot; independent questions; exact option-set match

One board, so its score

60.6

Multiplier for 1 of 10 boards

× 0.408

Index

60.6 × 0.408 = 24.7

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedPIC59.559.5×0.482, near its ceilingNear its ceiling: leader 80.7, weight 0.482zero-shot; independent questions; exact option-set match

One board, so its score

59.5

Multiplier for 1 of 10 boards

× 0.408

Index

59.5 × 0.408 = 24.3

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• EHR-Complex0.59059.0×0.875Table 3; macro-average across 12 intent/scope columns; temperature 0; up to 50 SQL/Python interaction turns

One board, so its score

59.0

Multiplier for 1 of 10 boards

× 0.408

Index

59.0 × 0.408 = 24.1

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedPIC59.159.1×0.482, near its ceilingNear its ceiling: leader 80.7, weight 0.482zero-shot; independent questions; exact option-set match

One board, so its score

59.1

Multiplier for 1 of 10 boards

× 0.408

Index

59.1 × 0.408 = 24.1

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)69.1469.1×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID openai/o4-mini-2025-04-16; max_output_tokens=30000; reasoning_effort=high
• MedCode (Vals AI)33.7933.8×0.911model ID openai/o4-mini-2025-04-16; reasoning_effort=high; max_output_tokens=30000

Weighted average of 2 boards

41.4

Multiplier for 2 of 10 boards

× 0.577

Index

41.4 × 0.577 = 23.9

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)77.1377.1×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID alibaba/qwen3-vl-plus-2025-09-23; temperature=1; max_output_tokens=30000
• MedCode (Vals AI)31.6531.7×0.911model ID alibaba/qwen3-vl-plus-2025-09-23; temperature=1; max_output_tokens=30000

Weighted average of 2 boards

41.5

Multiplier for 2 of 10 boards

× 0.577

Index

41.5 × 0.577 = 23.9

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)67.7367.7×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID mistralai/mistral-medium-3.5; temperature=1; top_p=0.95; max_output_tokens=30000; reasoning_effort=high
• MedCode (Vals AI)33.7533.8×0.911model ID mistralai/mistral-medium-3.5; reasoning_effort=high; temperature=1; top_p=0.95; max_output_tokens=30000

Weighted average of 2 boards

41.1

Multiplier for 2 of 10 boards

× 0.577

Index

41.1 × 0.577 = 23.7

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)70.6270.6×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID alibaba/qwen3.5-flash; temperature=1; max_output_tokens=30000
• MedCode (Vals AI)33.0033.0×0.911model ID alibaba/qwen3.5-flash; temperature=1; max_output_tokens=30000

Weighted average of 2 boards

41.1

Multiplier for 2 of 10 boards

× 0.577

Index

41.1 × 0.577 = 23.7

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedXpertQA (MM)58.158.1×0.463, near its ceilingNear its ceiling: leader 81.5, weight 0.463Google model card; MedXPertQA MM row; vendor-reported; protocol differs from other source families

One board, so its score

58.1

Multiplier for 1 of 10 boards

× 0.408

Index

58.1 × 0.408 = 23.7

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)79.7279.7×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID grok/grok-4-fast-non-reasoning; temperature=1; top_p=0.95; max_output_tokens=30000
• MedCode (Vals AI)30.0430.0×0.911model ID grok/grok-4-fast-non-reasoning; temperature=1; top_p=0.95; max_output_tokens=30000

Weighted average of 2 boards

40.7

Multiplier for 2 of 10 boards

× 0.577

Index

40.7 × 0.577 = 23.5

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)68.6368.6×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID zai/glm-4.7; temperature=1; top_p=1; max_output_tokens=30000
• MedCode (Vals AI)32.7732.8×0.911model ID zai/glm-4.7; temperature=1; top_p=1; max_output_tokens=30000

Weighted average of 2 boards

40.5

Multiplier for 2 of 10 boards

× 0.577

Index

40.5 × 0.577 = 23.4

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)72.1572.2×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID xiaomi/mimo-v2.5; temperature=1; top_p=0.95; max_output_tokens=30000
• MedCode (Vals AI)31.8931.9×0.911model ID xiaomi/mimo-v2.5; temperature=1; top_p=0.95; max_output_tokens=30000

Weighted average of 2 boards

40.6

Multiplier for 2 of 10 boards

× 0.577

Index

40.6 × 0.577 = 23.4

Score: the result on the board’s own 0 to 100 scale. Dot: this model; tick: the board’s leader. Weight: how much the board counts in the weighted average; bold where the board is near its ceiling and counts for less. Each result links to its quote and page; the line under it is the setting it was measured with.

• MedScribe (Vals AI)83.8583.9×0.25, near its ceilingNear its ceiling: leader 91.4, weight 0.25model ID alibaba/qwen3.8-27b; temperature=1; top_p=0.95; max_output_tokens=30000; reasoning_effort=xhigh
• MedCode (Vals AI)28.7028.7×0.911model ID alibaba/qwen3.8-27b; reasoning_effort=xhigh; temperature=1; top_p=0.95; max_output_tokens=30000

Weighted average of 2 boards

40.6

Multiplier for 2 of 10 boards

× 0.577

Index

40.6 × 0.577 = 23.4