Benchmarks

Compare AI models

Pick two models to see release dates, benchmark scores, and community ratings side by side.

Frontier benchmarks

Frontier board

6 releases. Five ways to compare.

Selected for recent releases and benchmark coverage, not a single overall rank. Each column sorts the same models by its own score. No reading means no bar.

○ Independent evaluation · ◆ Vendor or self-reported, labeled per reading.

Coding

BenchLM · 0 to 100

  1. Gemini 4 Neon92.9Independent

    Google DeepMind

  2. GPT-6.2 Astra91.9Independent

    OpenAI

  3. Gemini 4 Carbon91.8Independent

    Google DeepMind

  4. Claude Opus 5.588.5Independent

    Anthropic

  5. GPT-6.2 Sol83.9Independent

    OpenAI

  6. GPT-6 Astra82.2Independent

    OpenAI

Agentic

BenchLM · 0 to 100

  1. GPT-6.2 Astra92.9Independent

    OpenAI

  2. Gemini 4 Neon91.9Independent

    Google DeepMind

  3. Gemini 4 Carbon88.6Independent

    Google DeepMind

  4. GPT-6 Astra87.3Independent

    OpenAI

  5. Claude Opus 5.587.1Independent

    Anthropic

  6. GPT-6.2 Sol86.6Independent

    OpenAI

Reasoning

BenchLM · 0 to 100

  1. Gemini 4 Neon88.9Independent

    Google DeepMind

  2. Gemini 4 Carbon87.8Independent

    Google DeepMind

  3. GPT-6.2 Astra87.3Independent

    OpenAI

  4. GPT-6 Astra87Independent

    OpenAI

  5. GPT-6.2 Sol81.5Independent

    OpenAI

  6. Claude Opus 5.581Independent

    Anthropic

Knowledge

BenchLM · 0 to 100

  1. GPT-6.2 Astra88.8Independent

    OpenAI

  2. Gemini 4 Neon87.9Independent

    Google DeepMind

  3. Gemini 4 Carbon83.4Independent

    Google DeepMind

  4. GPT-6.2 Sol81.4Independent

    OpenAI

  5. GPT-6 Astra80Independent

    OpenAI

  6. Claude Opus 5.571.7Independent

    Anthropic

Arena

LM Arena · Elo

  1. Anthropic

  2. GPT-6 Astra1,476.48Independent

    OpenAI

  3. Gemini 4 CarbonNot reported

    Google DeepMind

  4. Gemini 4 NeonNot reported

    Google DeepMind

  5. GPT-6.2 AstraNot reported

    OpenAI

  6. GPT-6.2 SolNot reported

    OpenAI

Score links open the source; hover for capture details. Bar lengths start at zero. Arena lengths use the full public board maximum.

The shape of the scores

First three featured models. Toggle a name to isolate a line. Missing readings leave gaps, not zeroes.

CodingAgenticReasoningKnowledgeArena050100Gemini 4 Neon, Coding: 92.9. Independent. Captured 2026-11-22.Gemini 4 Neon, Agentic: 91.9. Independent. Captured 2026-11-22.Gemini 4 Neon, Reasoning: 88.9. Independent. Captured 2026-11-22.Gemini 4 Neon, Knowledge: 87.9. Independent. Captured 2026-11-22.GPT-6.2 Astra, Coding: 91.9. Independent. Captured 2026-11-22.GPT-6.2 Astra, Agentic: 92.9. Independent. Captured 2026-11-22.GPT-6.2 Astra, Reasoning: 87.3. Independent. Captured 2026-11-22.GPT-6.2 Astra, Knowledge: 88.8. Independent. Captured 2026-11-22.Gemini 4 Carbon, Coding: 91.8. Independent. Captured 2026-11-22.Gemini 4 Carbon, Agentic: 88.6. Independent. Captured 2026-11-22.Gemini 4 Carbon, Reasoning: 87.8. Independent. Captured 2026-11-22.Gemini 4 Carbon, Knowledge: 83.4. Independent. Captured 2026-11-22.

3 of 3 series shown.

Axes share drawing space, not meaning. Arena is Elo divided by the current board maximum; it is not a percentage accuracy.

The newest arrival

GPT-6.2 Sol

Released 2026-11-20 · At least two readings

No predecessor is linked in the catalog yet. We do not infer a generation jump from a name.

Latest release with linked lineage

Claude Fable 5.1 vs Claude Fable 5

Coding
+3.4pts
78 to 81.4 Independent2026-11-22 / 2026-11-22
Agentic
+8.5pts
72.1 to 80.6 Independent2026-11-22 / 2026-11-22
Reasoning
+4.2pts
71 to 75.2 Independent2026-11-22 / 2026-11-22
Knowledge
+0.2pts
77 to 77.2 Independent2026-11-22 / 2026-11-22
Arena
-0.09Elo
1,503.28 to 1,503.19 Independent2026-10-26 / 2026-10-26

Latest snapshots, not a controlled before-and-after trial. Capture dates and evaluation setups may differ.

Scores and sources

Frontier board: original scores. Not reported is missing, not zero.
ModelMetricScore / provenanceCapturedSource label / votes
Gemini 4 NeonCodingBenchLM.ai92.9Independent2026-11-22Gemini 4 Neon
GPT-6.2 AstraCodingBenchLM.ai91.9Independent2026-11-22GPT-6.2 Astra
Gemini 4 CarbonCodingBenchLM.ai91.8Independent2026-11-22Gemini 4 Carbon
Claude Opus 5.5CodingBenchLM.ai88.5Independent2026-11-22Claude Opus 5.5
GPT-6.2 SolCodingBenchLM.ai83.9Independent2026-11-22GPT-6.2 Sol
GPT-6 AstraCodingBenchLM.ai82.2Independent2026-11-22GPT-6 Astra
GPT-6.2 AstraAgenticBenchLM.ai92.9Independent2026-11-22GPT-6.2 Astra
Gemini 4 NeonAgenticBenchLM.ai91.9Independent2026-11-22Gemini 4 Neon
Gemini 4 CarbonAgenticBenchLM.ai88.6Independent2026-11-22Gemini 4 Carbon
GPT-6 AstraAgenticBenchLM.ai87.3Independent2026-11-22GPT-6 Astra
Claude Opus 5.5AgenticBenchLM.ai87.1Independent2026-11-22Claude Opus 5.5
GPT-6.2 SolAgenticBenchLM.ai86.6Independent2026-11-22GPT-6.2 Sol
Gemini 4 NeonReasoningBenchLM.ai88.9Independent2026-11-22Gemini 4 Neon
Gemini 4 CarbonReasoningBenchLM.ai87.8Independent2026-11-22Gemini 4 Carbon
GPT-6.2 AstraReasoningBenchLM.ai87.3Independent2026-11-22GPT-6.2 Astra
GPT-6 AstraReasoningBenchLM.ai87Independent2026-11-22GPT-6 Astra
GPT-6.2 SolReasoningBenchLM.ai81.5Independent2026-11-22GPT-6.2 Sol
Claude Opus 5.5ReasoningBenchLM.ai81Independent2026-11-22Claude Opus 5.5
GPT-6.2 AstraKnowledgeBenchLM.ai88.8Independent2026-11-22GPT-6.2 Astra
Gemini 4 NeonKnowledgeBenchLM.ai87.9Independent2026-11-22Gemini 4 Neon
Gemini 4 CarbonKnowledgeBenchLM.ai83.4Independent2026-11-22Gemini 4 Carbon
GPT-6.2 SolKnowledgeBenchLM.ai81.4Independent2026-11-22GPT-6.2 Sol
GPT-6 AstraKnowledgeBenchLM.ai80Independent2026-11-22GPT-6 Astra
Claude Opus 5.5KnowledgeBenchLM.ai71.7Independent2026-11-22Claude Opus 5.5
Claude Opus 5.5ArenaLM Arena1,508.73Independent2026-10-26Claude Opus 5.5 (High)10,322 votes
GPT-6 AstraArenaLM Arena1,476.48Independent2026-10-26GPT-6 Astra (Max)14,076 votes
Gemini 4 CarbonArenaLM ArenaNot reportedNot reportedNot reported
Gemini 4 NeonArenaLM ArenaNot reportedNot reportedNot reported
GPT-6.2 AstraArenaLM ArenaNot reportedNot reportedNot reported
GPT-6.2 SolArenaLM ArenaNot reportedNot reportedNot reported
Additional readings for these models. Separate benchmarks, not substitutes for the five chart axes.
ModelMetric / sourceScore / provenanceCaptured
Gemini 4 NeonMathBenchLM.ai83.9Independent2026-11-22
Gemini 4 NeonMultimodalGroundedBenchLM.ai80.9Independent2026-11-22
Gemini 4 NeonInstructionFollowingBenchLM.ai86.9Independent2026-11-22
GPT-6.2 AstraMathBenchLM.ai83.5Independent2026-11-22
GPT-6.2 AstraInstructionFollowingBenchLM.ai83.3Independent2026-11-22
GPT-6.2 AstraMultimodalGroundedBenchLM.ai90.3Independent2026-11-22
Gemini 4 CarbonMathBenchLM.ai80.4Independent2026-11-22
Gemini 4 CarbonMultimodalGroundedBenchLM.ai85.9Independent2026-11-22
Gemini 4 CarbonInstructionFollowingBenchLM.ai82.6Independent2026-11-22
GPT-6 AstraHumanity's Last ExamEpoch AI54.8Independent2026-11-22
GPT-6 AstraMultimodalGroundedBenchLM.ai79.2Independent2026-11-22
GPT-6 AstraMathBenchLM.ai79.6Independent2026-11-22
GPT-6 AstraMultilingualBenchLM.ai74.1Independent2026-11-22
GPT-6 AstraInstructionFollowingBenchLM.ai78.6Independent2026-11-22
Claude Opus 5.5MultimodalGroundedBenchLM.ai80.6Independent2026-11-22
Claude Opus 5.5GDP.pdfEpoch AI48.8Independent2026-11-22
Claude Opus 5.5MathBenchLM.ai77.5Independent2026-11-22
Claude Opus 5.5MultilingualBenchLM.ai74.4Independent2026-11-22
Claude Opus 5.5InstructionFollowingBenchLM.ai76.2Independent2026-11-22
GPT-6.2 SolInstructionFollowingBenchLM.ai74.4Independent2026-11-22
GPT-6.2 SolMultimodalGroundedBenchLM.ai84.2Independent2026-11-22
Open page →
Provider
OpenAI
-
Out yet?
No
-
Status
Speculated
-
Release date
-
-
Vibe rating?OutYet's community rating: signed-in users score the model 1–10. Shown as the average and the number of votes.
-
-
Predecessor
-
-
Successor
-
-