BenchLeader

GPT-6.2 Sol added, Astra for Law 6.2 added (17 Nov 2026)

GPT-6.2 Sol and Astra for Law 6.2 join the board. 3 new models, 6 top-10 rank changes, 24 price changes, 165 new benchmark results. Sources linked.

BenchLeader refreshed 1736 models from 27 of 29 data feeds (benchmark publishers, aggregators and price lists). 3 new models, 6 top-10 rank changes, 24 price changes, 165 new or updated benchmark results across 27 data feeds.

New models

  • GPT-6.2 Sol (OpenAI) appeared with 1 benchmark result. (details)
  • Astra for Law 6.2 (max) (OpenAI) is now listed at $16.00/M blended; no independent results yet. (details)
  • GPT-6.2 Astra (OpenAI) is now listed at $16.00/M blended; no independent results yet. (details)

Movement in the top 10

  • Claude Opus 5.5 (max) moved from #5 to #4 in the BenchLeader Index. (details)
  • Claude Fable 5.2 (high) moved from #4 to #5 in the BenchLeader Index. (details)
  • GPT-6 Astra (max) moved from #6 to #7 in the BenchLeader Index. (details)
  • GPT-6 Astra moved from #9 to #8 in the BenchLeader Index. (details)
  • Claude Fable 5.1 (max) moved from #7 to #9 in the BenchLeader Index. (details)
  • GPT-6.1 Sol (max) moved from #8 to #11 in the BenchLeader Index. (details)

Price changes

  • Qwen4 Max (max) cached input price moved from $0.075/M to $0.071/M (5% down). (details)
  • Muse Glimmer input price moved from $0.30/M to $0.35/M (17% up). (details)
  • Muse Glimmer output price moved from $1.20/M to $1.50/M (25% up). (details)
  • Qwen3.5-9B (thinking) cached input price moved from $0.10/M to $0.060/M (40% down). (details)
  • Qwen3.5-9B (no reasoning) cached input price moved from $0.10/M to $0.060/M (40% down). (details)
  • Mistral Medium 3.1 cached input price moved from $0.040/M to $0.044/M (10% up). (details)
  • gpt-oss-20b (low) cached input price moved from $0.037/M to $0.015/M (60% down). (details)
  • gpt-oss-20b (high) cached input price moved from $0.037/M to $0.015/M (60% down). (details)
  • gpt-oss-20b cached input price moved from $0.037/M to $0.015/M (60% down). (details)
  • Mistral Medium 3 cached input price moved from $0.040/M to $0.044/M (10% up). (details)
  • Qwen3.5-9B cached input price moved from $0.10/M to $0.060/M (40% down). (details)
  • gpt-oss-20b (medium) cached input price moved from $0.037/M to $0.015/M (60% down). (details)
  • GLM 5.3 Prime input price moved from $2.80/M to $1.68/M (40% down). (details)
  • GLM 5.3 Prime output price moved from $8.80/M to $5.28/M (40% down). (details)
  • GLM 5.3 Prime cached input price moved from $0.56/M to $0.34/M (40% down). (details)
  • Hy4 flash input price moved from $0.035/M to $0.070/M (100% up). (details)
  • Hy4 flash output price moved from $0.14/M to $0.28/M (100% up). (details)
  • Hy4 flash cached input price moved from $0.0015/M to $0.0030/M (100% up). (details)
  • Muse Glimmer (low) input price moved from $0.30/M to $0.35/M (17% up). (details)
  • Muse Glimmer (low) output price moved from $1.20/M to $1.50/M (25% up). (details)
  • Muse Glimmer (medium) input price moved from $0.30/M to $0.35/M (17% up). (details)
  • Muse Glimmer (medium) output price moved from $1.20/M to $1.50/M (25% up). (details)
  • Muse Glimmer (xhigh) input price moved from $0.30/M to $0.35/M (17% up). (details)
  • Muse Glimmer (xhigh) output price moved from $1.20/M to $1.50/M (25% up). (details)

New benchmark results

  • GDP.pdf: 8 new results, including Gemini 4 Carbon, Muse Spark 1.3, Kimi K3.1 Pro… (board)
  • GBAEval: 5 new results, including GPT-6 Astra, Claude Fable 5.2, Gemini 4 Argon… (board)
  • CursorBench: 4 new results, including Gemini 4 Hydrogen, Gemini 4 Hydrogen, Gemini 4 Hydrogen… (board)
  • ForecastBench: 3 new results, including Claude Opus 5.5, GPT-6 Sol, GPT-6 Luna (board)
  • Robo-bench: 3 new results, including Kimi K3.1 Pro, Qwen3.8 Max, DeepSeek V4.1 Flash 1017 (board)
  • Design Arena Agentic: 3 new results, including Muse Spark Open, Claude Haiku 5.5, Muse Glimmer (board)
  • ARC-AGI-1: 3 new results, including Mercury 3, Mercury 3, Mercury 3 (board)
  • ARC-AGI-2: 3 new results, including Mercury 3, Mercury 3, Mercury 3 (board)
  • PostTrainBench: 2 new results, including Claude Fable 5.2, Gemini 4 Carbon (board)
  • ALE-Bench: 2 new results, including Gemini 4 Carbon, Gemini 4 Hydrogen (board)
  • Vending-Bench 2: 2 new results, including Claude Fable 5.2, Gemini 4 Hydrogen (board)
  • FrontierSWE: 2 new results, including Gemini 4 Hydrogen, Inkling 2 Small (board)

Score revisions

  • Claude Opus 5.5 (max) on GDP.pdf: 30.6% → 48.8%. (details)
  • GPT-6 Astra (max) on GDP.pdf: 34.2% → 50.4%. (details)
  • GPT-6 Astra (max) on Design Arena Website: 1338 → 1346. (details)
  • Claude Fable 5.1 (max) on GDP.pdf: 27.6% → 49.4%. (details)
  • GPT-6.1 Sol (max) on GDP.pdf: 35% → 51.6%. (details)
  • Claude Fable 5.2 on Design Arena Agentic: 1370 → 1365. (details)
  • Gemini 4 Argon (high) on GDP.pdf: 28.4% → 53.6%. (details)
  • Gemini 4 Argon on Design Arena Website: 1316 → 1321. (details)
  • Claude Sonnet 5.5 (max) on GDP.pdf: 31.4% → 51.2%. (details)
  • Claude Opus 5 on Design Arena Website: 1304 → 1298. (details)
  • Muse Spark 1.1 on Design Arena Website: 1263 → 1272. (details)
  • Gemini 3.7 Flash (high) on ForecastBench: 66.2% → 65.7%. (details)
  • Kimi K3.1 Pro on Design Arena Agentic: 1300 → 1294. (details)
  • GPT-5.5 on Design Arena Website: 1248 → 1253. (details)
  • Muse Wonder (max) on GDP.pdf: 30.8% → 54.8%. (details)
  • Muse Wonder (max) on Design Arena Website: 1379 → 1373. (details)
  • Kimi K3 (max) on GDP.pdf: 19% → 39.8%. (details)
  • Gemini 3.8 Flash (high) on GDP.pdf: 23.2% → 44.2%. (details)
  • Qwen3.8 Max (max) on Design Arena Website: 1269 → 1276. (details)
  • Muse Spark 1.3 (xhigh) on Design Arena Website: 1360 → 1351. (details)
  • …and 56 more.

Speed changes

  • Gemini 4 Carbon time to first token changed from 13.14 s to 7.04 s. (details)
  • Claude Fable 5.1 time to first answer changed from 147.58 s to 205.02 s. (details)
  • GPT-6.1 Sol time to first token changed from 8.42 s to 3.16 s. (details)
  • GPT-5.6 Sol time to first token changed from 3.71 s to 2.38 s. (details)
  • Gemini 4 Argon output speed changed from 24 tok/s to 33 tok/s. (details)
  • Muse Spark 1.3 output speed changed from 91 tok/s to 150 tok/s. (details)
  • Grok 4.6 time to first token changed from 3.61 s to 1.91 s. (details)
  • Claude Opus 4.7 time to first answer changed from 14.54 s to 20.82 s. (details)
  • Grok 4.5 time to first token changed from 3.21 s to 1.36 s. (details)
  • Grok 4.20 Multi-Agent time to first token changed from 4.71 s to 8.00 s. (details)
  • GPT-5.2-Codex time to first token changed from 2.98 s to 1.78 s. (details)
  • MiMo-V2.6-Pro thinking time changed from 35.74 s to 23.17 s. (details)
  • DeepSeek V4 Flash output speed changed from 134 tok/s to 86 tok/s. (details)
  • Grok 4.7 time to first token changed from 5.79 s to 2.82 s. (details)
  • DeepSeek V3.2 time to first token changed from 2.26 s to 1.26 s. (details)
  • …and 85 more.

Source status

  • Source scale failed to update: degraded: scale_audiomc_audio, scale_audiomc_text, scale_prbench_finance, scale_prbench_legal returned nothing for a second run in a row; 126 rows carried forward from 2026-11-16.
  • Source designarena failed to update: degraded: returned 92 models, down 47% from 175; 92 rows carried forward from 2026-11-16.

Today's top five

  1. Claude Fable 5.2 (max) — 70.2
  2. Gemini 4 Carbon (high) — 68.6
  3. GPT-6.2 Astra (max) — 68.3
  4. Gemini 4 Neon (high) — 68.0
  5. GPT-6 Astra — 68.0

More: New AI model releases this week, and the full LLM leaderboard 2026.

This digest is generated automatically from the day's data changes. Every line links to the page where you can check the numbers and their source.

Head to head

Get this in your inbox

No ads, no tracking pixels, unsubscribe in one click.
What to receive