Benchmark Almanac

Bedside Bench

Doximity Ask V7 95.4% tops its leaderboard Claude Opus 5 94.6% leads the models we track
Not in model cards

Models answer 500 physician-written clinical cases on diagnosis, workup, escalation and medication safety. An LLM judge grades each answer against the authors' criteria.

Ten suites, with the headline as the macro average of suite means. 16 models in the Fireworks index. No maker page with details found.

Made by
Doximity
Kind
Medicine
Level
Expert
Scored
Graded by a model
Answers
Open-ended
Tasks
Kept private
Strongest model tested
GPT-6 Astra #3 on the capability index
Last model added
October 2026
Scores refreshed
November 12th, 2026
18
Entries on its own leaderboard, 17 models we track

Doximity Ask V7 tops the leaderboard with 95.4%

Best score per entry from the benchmark's own leaderboard, read on November 12th, 2026. Where a model ran with several settings or harnesses, we keep its best. Every run is in the next chart. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. The best model we track is Claude Opus 5, at 94.6%.

18 entries. Scroll sideways for the rest.

  1. 95.4 Doximity Ask V7
  2. 94.6 Claude Opus 5
  3. 94.3 Muse Spark Open
  4. 93.9 GPT-6.1 Sol
  5. 93.4 GPT-6 Astra
  6. 92.7 Gemini 4 Argon
  7. 91.5 Kimi K3.1 Pro
  8. 91.1 GPT-5.6 Sol
  9. 89.7 Kimi K3
  10. 89.1 Ember-1
  11. 88.1 GPT-6 Sol
  12. 87.2 GPT-6 Luna
  13. 86.9 GLM-5.3
  14. 86.5 Gemini 3.8 Flash
  15. 85.7 DeepSeek V4.1 Flash 1017
  16. 83.3 Grok 4.6
  17. 82.9 DeepSeek V4.1 Flash
  18. 80.6 DeepSeek V4 Pro 0813
  1. 1 Doximity Ask V7 95.4%
  2. 2 Claude Opus 5 94.6%
  3. 3 Muse Spark Open 94.3%
  4. 4 GPT-6.1 Sol 93.9%
  5. 5 GPT-6 Astra 93.4%
  6. 6 Gemini 4 Argon 92.7%
  7. 7 Kimi K3.1 Pro 91.5%
  8. 8 GPT-5.6 Sol 91.1%
  9. 9 Kimi K3 89.7%
  10. 10 Ember-1 89.1%
  11. 11 GPT-6 Sol 88.1%
  12. 12 GPT-6 Luna 87.2%
  13. 13 GLM-5.3 86.9%
  14. 14 Gemini 3.8 Flash 86.5%
  15. 15 DeepSeek V4.1 Flash 1017 85.7%
  16. 16 Grok 4.6 83.3%
  17. 17 DeepSeek V4.1 Flash 82.9%
  18. 18 DeepSeek V4 Pro 0813 80.6%

Source: Official Bedside Bench leaderboard

#EntryScoreSettingDateSource
1 Doximity Ask V7 95.4% Nov 2026 Official leaderboard
2 Claude Opus 5 94.6% Jul 2026 Official leaderboard
3 Muse Spark Open 94.3% Oct 2026 Official leaderboard
4 GPT-6.1 Sol 93.9% Sep 2026 Official leaderboard
5 GPT-6 Astra 93.4% Sep 2026 Official leaderboard
6 Gemini 4 Argon 92.7% Sep 2026 Official leaderboard
7 Kimi K3.1 Pro 91.5% Oct 2026 Official leaderboard
8 GPT-5.6 Sol 91.1% Jul 2026 Official leaderboard
9 Kimi K3 89.7% Jul 2026 Official leaderboard
10 Ember-1 89.1% Sep 2026 Official leaderboard
11 GPT-6 Sol 88.1% Sep 2026 Official leaderboard
12 GPT-6 Luna 87.2% Sep 2026 Official leaderboard
13 GLM-5.3 86.9% Aug 2026 Official leaderboard
14 Gemini 3.8 Flash 86.5% Sep 2026 Official leaderboard
15 DeepSeek V4.1 Flash 1017 85.7% Oct 2026 Official leaderboard
16 Grok 4.6 83.3% Aug 2026 Official leaderboard
17 DeepSeek V4.1 Flash 82.9% Sep 2026 Official leaderboard
18 DeepSeek V4 Pro 0813 80.6% Aug 2026 Official leaderboard

Score vs. cost per task

The line joins the runs that no other run beats on both measures. We redrew this chart from the numbers on the official leaderboard.

80% 90% 100% $0$0.03$0.05$0.07$0.1$0.13 Cost per task ($), lower is better Score GPT-5.6 Sol: Score 91.1%, cost per task $0.05 Kimi K3: Score 89.7%, cost per task $0.05 Grok 4.6: Score 83.3%, cost per task $0.01 DeepSeek V4 Pro 0813: Score 80.6%, cost per task $0.02 GLM-5.3: Score 86.9%, cost per task $0.02 Gemini 3.8 Flash: Score 86.5%, cost per task $0.009 GPT-6 Astra: Score 93.4%, cost per task $0.11 DeepSeek V4.1 Flash: Score 82.9%, cost per task $0.008 Ember-1: Score 89.1%, cost per task $0.04 GPT-6.1 Sol: Score 93.9%, cost per task $0.04 Gemini 4 Argon: Score 92.7%, cost per task $0.03 DeepSeek V4.1 Flash 1017: Score 85.7%, cost per task $0.007 Kimi K3.1 Pro: Score 91.5%, cost per task $0.05 Claude Opus 5: Score 94.6%, cost per task $0.12 GPT-6 Luna: Score 87.2%, cost per task $0.001 GPT-6 Sol: Score 88.1%, cost per task $0.01 Muse Spark Open: Score 94.3%, cost per task $0.02 Claude Opus 5Muse Spark OpenGPT-6 SolGPT-6 Luna 80% 90% 100% $0$0.03$0.05$0.07$0.1$0.13 Cost per task ($), lower is better Score GPT-5.6 Sol: Score 91.1%, cost per task $0.05 Kimi K3: Score 89.7%, cost per task $0.05 Grok 4.6: Score 83.3%, cost per task $0.01 DeepSeek V4 Pro 0813: Score 80.6%, cost per task $0.02 GLM-5.3: Score 86.9%, cost per task $0.02 Gemini 3.8 Flash: Score 86.5%, cost per task $0.009 GPT-6 Astra: Score 93.4%, cost per task $0.11 DeepSeek V4.1 Flash: Score 82.9%, cost per task $0.008 Ember-1: Score 89.1%, cost per task $0.04 GPT-6.1 Sol: Score 93.9%, cost per task $0.04 Gemini 4 Argon: Score 92.7%, cost per task $0.03 DeepSeek V4.1 Flash 1017: Score 85.7%, cost per task $0.007 Kimi K3.1 Pro: Score 91.5%, cost per task $0.05 Claude Opus 5: Score 94.6%, cost per task $0.12 GPT-6 Luna: Score 87.2%, cost per task $0.001 GPT-6 Sol: Score 88.1%, cost per task $0.01 Muse Spark Open: Score 94.3%, cost per task $0.02 Claude Opus 5Muse Spark OpenGPT-6 SolGPT-6 Luna
ModelSettingScoreCost per task
Claude Opus 5 94.6% $0.12
Muse Spark Open 94.3% $0.02
GPT-6.1 Sol 93.9% $0.04
GPT-6 Astra 93.4% $0.11
Gemini 4 Argon 92.7% $0.03
Kimi K3.1 Pro 91.5% $0.05
GPT-5.6 Sol 91.1% $0.05
Kimi K3 89.7% $0.05
Ember-1 89.1% $0.04
GPT-6 Sol 88.1% $0.01
GPT-6 Luna 87.2% $0
GLM-5.3 86.9% $0.02
Gemini 3.8 Flash 86.5% $0.01
DeepSeek V4.1 Flash 1017 85.7% $0.01
Grok 4.6 83.3% $0.01
DeepSeek V4.1 Flash 82.9% $0.01
DeepSeek V4 Pro 0813 80.6% $0.02

The best score, 94.6%, costs $0.12 a task

Runs on the cost frontier score higher than any run that's cheaper. Muse Spark Open comes within 0.3 points of the best for $0.02 a task, 7 times cheaper. The costs are the leaderboard's own estimates, so compare them only within this benchmark.

Anthropic OpenAI Meta Cost frontier One model at each effort level
0% 20% 40% 60% 80% 100% $0.001 $0.01 $0.10 $1.00 Cost per task, log scale Claude Opus 5 Muse Spark Open GPT-6 Sol GPT-6 Luna 0% 20% 40% 60% 80% 100% $0.001 $0.01 $0.10 $1.00 Cost per task, log scale Claude Opus 5 Muse Spark Open

Source: Official Bedside Bench leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Claude Opus 5 94.6% $0.12 4,699
Muse Spark Open 94.3% $0.02 3,914
GPT-6.1 Sol 93.9% $0.04 3,728
GPT-6 Astra 93.4% $0.11 2,087
Gemini 4 Argon 92.7% $0.03 2,858
Kimi K3.1 Pro 91.5% $0.05 4,575
GPT-5.6 Sol 91.1% $0.05 2,471
Kimi K3 89.7% $0.05 3,479
Ember-1 89.1% $0.04 2,668
GPT-6 Sol 88.1% $0.01 1,459
GPT-6 Luna 87.2% $0.001 2,863
GLM-5.3 86.9% $0.02 4,728
Gemini 3.8 Flash 86.5% $0.009 2,412
DeepSeek V4.1 Flash 1017 85.7% $0.007 5,421
Grok 4.6 83.3% $0.01 1,856
DeepSeek V4.1 Flash 82.9% $0.008 6,923
DeepSeek V4 Pro 0813 80.6% $0.02 3,878

2 models have held number one on the official leaderboard since 2023

Lab claims count from the day the card came out, independent runs from the model’s release.

OpenAI Anthropic
Own leaderboard Passed 85% in Jul 2026, by GPT-5.6 Sol
Claude Opus 5 94.6%

Source: Official Bedside Bench leaderboard

GPT-5.6 Sol on the official leaderboard passed 85% in July 2026

All the scores we have, one dot each, so a model tested at several settings appears more than once.

On the benchmark’s own leaderboard Best so far
0% 20% 40% 60% 80% 100% 2027 Near the ceiling (85%) Claude Opus 5 0% 20% 40% 60% 80% 100% 2027 Near the ceiling (85%) Claude Opus 5

Source: Official Bedside Bench leaderboard

No lab model card reports it yet

We read the headline results of 522 launch posts from 52 labs, and none of them list this benchmark.

More in Law, finance and medicine

All 179 in Law, finance and medicine
  • MedXpertQA-MM In use Models answer expert-level medical exam questions that combine images with clinical records and test results. Scoring is accuracy. Qwen 3.8 Max 80.4%
  • HealthBench Professional In use 525 physician-authored tasks from real clinician chats (care consult, documentation, research) with physician-adjudicated rubrics. Claude Sonnet 5.5 69.2%
  • HealthBench In use 5,000 realistic health conversations graded against physician-written rubrics. A length-adjusted variant penalizes verbosity. GPT-5 67.2%
  • FinSearchComp In use Agents search the web to fetch live financial data, look up history, and run multi-step investigations, across 635 expert-written questions. Scored by answer accuracy. Ling-3.1-flash 81.7%
  • KBL In use Models answer Korean legal knowledge and reasoning tasks and Korean bar exam questions. It has 7 knowledge tasks, 4 reasoning tasks and 2,510 bar exam items. Solar Pro 4 77.5%
  • KorMedMCQA In use Models answer 7,469 multiple-choice questions from Korean licensing exams for doctors, nurses, pharmacists and dentists, held from 2012 to 2024. Solar Pro 4 93.2%