The honest view from inside technical leadership
LLM Leaderboard

LLM Leaderboard: November 2026

The models worth putting in front of your engineers, ranked on agentic coding, reasoning and human preference, with the price you'll pay. Every number links to the benchmark that published it.

#1

Claude Fable 5.2

Anthropic

95.5CTO Index

Terminal-Bench
65.8%
WebDev
1824
ARC-AGI-2
93.8%
Text Arena
1503
Price $/M
$8 / $40
#2

Claude Opus 5.5

Anthropic

95.3CTO Index

Terminal-Bench
65.8%
WebDev
1819
ARC-AGI-2
93.3%
Text Arena
1506
Price $/M
$4 / $20
#3

Gemini 4 Carbon

Google

92.4CTO Index

Terminal-Bench
64.5%
WebDev
1769
ARC-AGI-2
94.2%
Text Arena
1527
Price $/M
$2 / $10
45models tracked
39ranked
5primary-source benchmarks
22 November 2026data snapshot
Best for coding agentsCopilot Fusion71.2% Terminal-BenchBest valueGLM-5.3-Flash$0.02 / $0.07 per M tokensBest open weightsKimi K3.1 ProKimi K3.1 Pro License

The full leaderboard

Sort by any column. Click a model for its full profile and sources.

45 of 45 models
Model
#1Claude Fable 5.2Anthropic95.565.8%182493.8%150381.2%$8 / $406.01M
#2Claude Opus 5.5Anthropic95.365.8%181993.3%1506—$4 / $2011.91M
#3Gemini 4 CarbonGoogle92.464.5%176994.2%152765.1%$2 / $1023.1Not listed
#4GPT-6.2 AstraOpenAI88.062.1%176695.8%146870.3%$8 / $405.51.1M
#5Gemini 4 NeonGoogle87.661.2%1765—1513—$1.50 / $7.5029.21M
#6GPT-6 AstraOpenAI87.260.3%176495.0%147562.7%$10 / $504.41.1M
#7Claude Sonnet 5.5Anthropic87.163.9%177283.9%147324.3%$2 / $1021.81M
#8GPT-6.1 SolOpenAI86.460.9%174394.2%148352.7%$2 / $1021.61.1M
#9Claude Fable 5.1Anthropic84.957.6%174290.0%1503—$10 / $504.21M
#10GPT-6.2 SolOpenAI84.661.5%175485.8%146554.9%$2 / $1021.21.1M
#11Gemini 4 HydrogenGoogle81.456.4%170192.5%1490—$1.50 / $7.5027.1Not listed
#12Gemini 4 ArgonGoogle81.355.5%168291.3%152049.8%$2 / $1020.31M
#13Claude Opus 5Anthropic78.953.3%169290.4%149030.2%$5 / $257.91M
#14Kimi K3.1 ProMoonshot AIOpen weights77.759.7%1683—1482—$2 / $1019.41M
#15GPT-6 SolOpenAI74.350.3%167489.6%14584.6%$2 / $1018.61.1M
#16GPT-6.1 TerraOpenAI73.353.3%1685—1474—$1 / $536.71.1M
#17Claude Fable 5Anthropic71.444.8%162689.2%1504—$10 / $503.61M
#18Muse WonderMeta69.347.0%1647—1511—$4 / $208.71M
#19GPT-5.6 SolOpenAI67.438.5%161592.5%14857.8%$4 / $208.4Not listed
#20DeepSeek V4.1 FlashDeepSeekOpen weights66.138.5%165878.1%14750.2%$0.15 / $0.60252.01M
#21GLM-5.3Z.aiOpen weights62.041.2%1623—1482—$1.40 / $4.4028.91M
#22GLM-5.3-FlashZ.aiOpen weights60.237.0%161065.8%1475—$0.02 / $0.071853.71M
#23Grok 4.7xAI60.038.8%163961.4%14431.8%$2 / $620.0500K
#24Claude Haiku 5.5Anthropic57.140.0%159561.3%14210.5%$0.10 / $0.50285.51M
#25Qwen3.8-MaxAlibaba57.127.0%1674—1484—$2 / $619.01M
#26Gemini 3.8 FlashGoogle55.918.2%158489.2%149710.4%$0.75 / $3.7537.31M
#27Kimi K3Moonshot AIOpen weights55.623.0%165960.4%1488—$3 / $159.31M
#28MiMo-V2.6-ProXiaomiOpen weights53.227.0%1622—1481—$0.43 / $0.8798.51.1M
#29Claude Opus 4.8Anthropic52.823.6%155572.1%14821.5%$5 / $255.31M
#30Qwen3.8-27BAlibabaOpen weights52.3—159042.4%1435—$0.40 / $349.8262K
#31Gemini 3.7 FlashGoogle51.912.4%158984.6%1490—$0.75 / $3.7534.61M
#32Grok 4.6xAI51.419.7%161867.1%14572.1%$1.60 / $4.8021.4500K
#33Inkling 2 SmallThinking MachinesOpen weights50.113.0%164381.7%14120.3%$0.30 / $1105.5Not listed
#34Muse Spark 1.3Meta49.014.8%1653—1498—$1.25 / $4.2524.51M
#35Mistral Large 4MistralOpen weights46.627.6%1573—1423—$0.70 / $2.1044.41M
#36MiMo-V2.6-FlashXiaomiOpen weights45.316.1%1638—1448—$0.14 / $0.28258.61M
#37GPT-6 LunaOpenAI44.915.2%157859.3%14430.2%$0.10 / $0.50224.41.1M
#38Claude Sonnet 5Anthropic37.912.1%1539—1461—$2 / $109.51M
#39Beam 1.0Reflection AIOpen weights27.76.4%143723.5%1434—$0.30 / $158.31M
–Copilot FusionGitHubPartial100.071.2%————Not listed—Not listed
–Qwen4-MaxAlibabaPartial79.3—1751———$1.50 / $4.50—Not listed
–Muse Wonder FullMetaPartial71.7—1723———Not listed—Not listed
–Step 5 PreviewStepFunPartial55.5—1570—1449—$1 / $2.70—1M
–Hy4TencentPartial53.437.9%1650———Not listed—Not listed
–Muse EmberMetaPartial43.730.6%1608———$0.10 / $0.50—Not listed

Models need at least 3 of the 4 indexed benchmarks to get a rank; the rest are marked Partial. A dash means the source has not published a result for that model. ARC-AGI-3 is shown for reference and is not part of the index.

How the CTO Index works

Public leaderboards each measure one thing, and model vendors quote whichever one flatters them. The CTO Index puts the ones that predict real engineering value side by side, weighted toward what engineering teams actually buy models for.

  1. Primary sources only. Each score is copied from the benchmark's own leaderboard on the snapshot date, never from an aggregator or a vendor press release.
  2. Normalise. The best tracked model on each benchmark scores 100. Percentages scale linearly. Elo ratings become the expected head-to-head win rate against the leader, so a 100-point Elo gap is about 72.
  3. Weight. Terminal-Bench 4.1 35%, WebDev Arena 25%, ARC-AGI-2 20%, Text Arena 20%. Agentic coding gets the most weight because that is where most engineering spend on models is going.
  4. Require coverage. The weighted mean is taken over the benchmarks a model has. Models with fewer than 3 of the four are listed but not ranked.
  5. Value. Index points per blended dollar, using a 3:1 input-to-output token mix typical of coding and RAG workloads.

What this does not tell you: latency, rate limits, data residency or how a model handles your codebase. Use the index to pick a shortlist of two or three, then run them on twenty of your own real tasks.

Benchmarks we track

Head-to-head comparisons

Frequently asked questions

What is the best AI model in November 2026?

Claude Fable 5.2 from Anthropic leads the Diary of a CTO Index with 95.5 points, followed by Claude Opus 5.5 (95.3) and Gemini 4 Carbon (92.4). The top three are close enough that price, latency and your own task evaluations should decide between them.

Which LLM is best for coding agents?

On Terminal-Bench 4.1, which measures agents finishing real tasks in a terminal, Copilot Fusion has the top published score at 71.2%. Scores within about three points of each other are statistical ties.

What is the best value LLM?

GLM-5.3-Flash gives the most CTO Index points per blended dollar, at $0.02 / $0.07 per million input/output tokens. Value matters most for high-volume workloads such as classification, extraction and support.

What is the best open-weight model?

Kimi K3.1 Pro from Moonshot AI is the highest-ranked model whose weights you can download and self-host, released under the Kimi K3.1 Pro License.

Where does this leaderboard get its data?

Every score is taken from the benchmark's own public leaderboard: Terminal-Bench, ARC Prize and LMArena. We do not copy aggregator scores or run private evaluations. Prices and context windows are list prices as shown on LMArena on the snapshot date.

How is the CTO Index calculated?

Each benchmark is normalised so the best tracked model scores 100. Elo ratings become the expected win rate against the leader. We then take a weighted mean: Terminal-Bench 35%, WebDev Arena 25%, ARC-AGI-2 20%, Text Arena 20%. A model needs at least 3 of those four to be ranked.

Sources