BenchLeader
Menu
Best LLMsChoose a modelCoding plansLeaderboardsModelsBenchmarksProvidersReleases4

Who is winning the AI race?

Data as of 2 Nov 2026 · last updated · by BenchLeader

Claude Opus 5.5 by Anthropic is the best AI model available on the BenchLeader Index (69.2), in the lead since September 2026; the lead has changed hands 9 times since June 2024. OpenAI has held it longest, 20 months in all. The best open-weights model, Kimi K3 by Moonshot AI, scores 62.1: 7.1 points behind the best closed model, against 4.5 in June 2024.

Best model available from each lab, by monthClaude Opus 5.5 by Anthropic is the best AI model available on the BenchLeader Index (69.2), in the lead since September 2026; the lead has changed hands 9 times since June 2024.Best model available from each lab, by monthAnthropicGoogleOpenAIMetaAlibaba35455565Jul2025Jul2026JulBenchLeader Index, as scored todayAnthropic, Jun: 39Anthropic, Jul: 39Anthropic, Aug: 39Anthropic, Sep: 39Anthropic, Oct: 47.1Anthropic, Nov: 47.1Anthropic, Dec: 47.1Anthropic, Jan: 47.1Anthropic, Feb: 51.9Anthropic, Mar: 51.9Anthropic, Apr: 51.9Anthropic, May: 55.2Anthropic, Jun: 55.2Anthropic, Jul: 55.2Anthropic, Aug: 57.9Anthropic, Sep: 57.9Anthropic, Oct: 57.9Anthropic, Nov: 59.7Anthropic, Dec: 59.7Anthropic, Jan: 59.7Anthropic, Feb: 62.6Anthropic, Mar: 62.6Anthropic, Apr: 63.1Anthropic, May: 63.1Anthropic, Jun: 66.6Anthropic, Jul: 66.6Anthropic, Aug: 66.6Anthropic, Sep: 69.2Anthropic, Oct: 69.2Anthropic, Nov: 69.2Google, Sep: 41.5Google, Oct: 41.5Google, Nov: 41.5Google, Dec: 41.5Google, Jan: 41.5Google, Feb: 48.4Google, Mar: 54.5Google, Apr: 54.5Google, May: 54.5Google, Jun: 54.5Google, Jul: 54.5Google, Aug: 54.5Google, Sep: 54.5Google, Oct: 54.5Google, Nov: 60.2Google, Dec: 60.2Google, Jan: 60.2Google, Feb: 60.6Google, Mar: 60.6Google, Apr: 60.6Google, May: 62.5Google, Jun: 62.5Google, Jul: 62.5Google, Aug: 62.5Google, Sep: 65.9Google, Oct: 68.9Google, Nov: 68.9OpenAI, Jun: 43.3OpenAI, Jul: 43.3OpenAI, Aug: 43.3OpenAI, Sep: 43.3OpenAI, Oct: 43.3OpenAI, Nov: 43.3OpenAI, Dec: 53.4OpenAI, Jan: 53.4OpenAI, Feb: 53.4OpenAI, Mar: 53.4OpenAI, Apr: 61OpenAI, May: 61OpenAI, Jun: 61OpenAI, Jul: 61OpenAI, Aug: 61OpenAI, Sep: 61OpenAI, Oct: 61OpenAI, Nov: 61OpenAI, Dec: 61OpenAI, Jan: 61OpenAI, Feb: 61OpenAI, Mar: 63.3OpenAI, Apr: 64.4OpenAI, May: 64.4OpenAI, Jun: 64.4OpenAI, Jul: 65.4OpenAI, Aug: 65.4OpenAI, Sep: 67.2OpenAI, Oct: 67.2OpenAI, Nov: 67.2Meta, Jun: 38.8Meta, Jul: 41.9Meta, Aug: 41.9Meta, Sep: 41.9Meta, Oct: 41.9Meta, Nov: 41.9Meta, Dec: 41.9Meta, Jan: 41.9Meta, Feb: 41.9Meta, Mar: 41.9Meta, Apr: 42.2Meta, May: 42.2Meta, Jun: 42.2Meta, Jul: 42.2Meta, Aug: 42.2Meta, Sep: 42.2Meta, Oct: 42.2Meta, Nov: 42.2Meta, Dec: 42.2Meta, Jan: 42.2Meta, Feb: 42.2Meta, Mar: 42.2Meta, Apr: 64.3Meta, May: 64.3Meta, Jun: 64.3Meta, Jul: 64.3Meta, Aug: 64.3Meta, Sep: 64.3Meta, Oct: 64.3Meta, Nov: 64.3Alibaba, Sep: 43.2Alibaba, Oct: 43.2Alibaba, Nov: 43.2Alibaba, Dec: 43.2Alibaba, Jan: 43.2Alibaba, Feb: 43.2Alibaba, Mar: 43.2Alibaba, Apr: 50Alibaba, May: 50Alibaba, Jun: 50Alibaba, Jul: 51.9Alibaba, Aug: 51.9Alibaba, Sep: 54.3Alibaba, Oct: 54.3Alibaba, Nov: 54.3Alibaba, Dec: 54.3Alibaba, Jan: 54.3Alibaba, Feb: 54.3Alibaba, Mar: 59.1Alibaba, Apr: 62.4Alibaba, May: 62.4Alibaba, Jun: 62.4Alibaba, Jul: 62.4Alibaba, Aug: 62.4Alibaba, Sep: 62.4Alibaba, Oct: 62.4Alibaba, Nov: 62.4AnthropicGoogleOpenAIMetaAlibabaIndex: BenchLeader v1.4, as scored today · by release datedata as of 2 Nov 2026 · CC BY 4.0www.benchleader.com/stats/ai-race
Claude Opus 5.5 by Anthropic is the best AI model available on the BenchLeader Index (69.2), in the lead since September 2026; the lead has changed hands 9 times since June 2024. Each line is a lab's best model released by that month, on today's BenchLeader Index. Chart image (SVG)

Key statistics

Each statistic has its own link (#) to cite it directly.

  • Claude Opus 5.5 by Anthropic leads the BenchLeader Index at 69.2 and has been the best model available since September 2026 (BenchLeader, 2 Nov 2026). #
  • Read back from release dates on today's scores, the lead has changed hands 9 times since June 2024, among 3 labs: OpenAI (20 months), Anthropic (9 months) and Google (1 month) (BenchLeader, 2 Nov 2026). #
  • The longest run at the top was o3 by OpenAI: 10 months, April 2025 to January 2026 (BenchLeader, 2 Nov 2026). #
  • The best open-weights model, Kimi K3 by Moonshot AI, scores 62.1: 7.1 points behind the best closed model, against 4.5 in June 2024 (BenchLeader, 2 Nov 2026). #
  • No open-weights model has been the best available in any month since June 2024 (BenchLeader, 2 Nov 2026). #

Every lead since June 2024

The best model available each month, run by run, newest first. By months in the lead: OpenAI 20, Anthropic 9, Google 1.

In the leadModelIndex todayMonths
September 2026 to November 2026 (now)Claude Opus 5.5 Anthropic69.23
June 2026 to August 2026Claude Fable 5 Anthropic66.63
April 2026 to May 2026GPT-5.5 OpenAI64.42
March 2026GPT-5.4 OpenAI63.31
February 2026Claude Opus 4.6 Anthropic62.61
April 2025 to January 2026o3 OpenAI61.010
March 2025Gemini 2.5 Pro Google54.51
December 2024 to February 2025o1 OpenAI53.43
October 2024 to November 2024Claude 3.5 Sonnet Anthropic47.12
June 2024 to September 2024GPT-4o OpenAI43.34

Open weights against closed

The best open-weights model and the best closed model available each month. The gap between the lines is how far behind anyone running their own weights was.

Best open-weights vs best closed model, by monthClaude Opus 5.5 by Anthropic is the best AI model available on the BenchLeader Index (69.2), in the lead since September 2026; the lead has changed hands 9 times since June 2024. OpenAI has held it longest, 20 months in all. The best open-weights model, Kimi K3 by Moonshot AI, scores 62.1: 7.1 points behind the best closed model, against 4.5 in June 2024.Best open-weights vs best closed model, by monthClosedOpen weights35455565Jul2025Jul2026JulBenchLeader Index, as scored todayClosed, Jun: 43.3Closed, Jul: 43.3Closed, Aug: 43.3Closed, Sep: 43.3Closed, Oct: 47.1Closed, Nov: 47.1Closed, Dec: 53.4Closed, Jan: 53.4Closed, Feb: 53.4Closed, Mar: 54.5Closed, Apr: 61Closed, May: 61Closed, Jun: 61Closed, Jul: 61Closed, Aug: 61Closed, Sep: 61Closed, Oct: 61Closed, Nov: 61Closed, Dec: 61Closed, Jan: 61Closed, Feb: 62.6Closed, Mar: 63.3Closed, Apr: 64.4Closed, May: 64.4Closed, Jun: 66.6Closed, Jul: 66.6Closed, Aug: 66.6Closed, Sep: 69.2Closed, Oct: 69.2Closed, Nov: 69.2Open weights, Jun: 38.8Open weights, Jul: 41.9Open weights, Aug: 41.9Open weights, Sep: 43.2Open weights, Oct: 43.2Open weights, Nov: 43.2Open weights, Dec: 44.4Open weights, Jan: 47.3Open weights, Feb: 47.3Open weights, Mar: 47.3Open weights, Apr: 50Open weights, May: 50Open weights, Jun: 50Open weights, Jul: 51.9Open weights, Aug: 51.9Open weights, Sep: 52.3Open weights, Oct: 52.6Open weights, Nov: 53.3Open weights, Dec: 56.7Open weights, Jan: 58.7Open weights, Feb: 60.2Open weights, Mar: 60.2Open weights, Apr: 60.2Open weights, May: 60.2Open weights, Jun: 60.2Open weights, Jul: 62.1Open weights, Aug: 62.1Open weights, Sep: 62.1Open weights, Oct: 62.1Open weights, Nov: 62.1ClosedOpen weightsIndex: BenchLeader v1.4, as scored today · by release datedata as of 2 Nov 2026 · CC BY 4.0www.benchleader.com/stats/ai-race
The best open-weights model, Kimi K3 by Moonshot AI, scores 62.1: 7.1 points behind the best closed model, against 4.5 in June 2024. Chart image (SVG)

How these numbers are made

Read back from release dates: for each month, the models released by its end, scored on today's BenchLeader Index. It shows who was ahead as the field is measured now, not what was measured at the time. Only models with at least 10 independent results count, so a model scored on a handful of boards cannot take the lead, and models BenchLeader no longer tracks are not in it.

A lead of a point or two is within the Index's margin of error; see the methodology. The labs charted are the five whose best model scores highest today.

Use this chart

These counts are BenchLeader's own and free to reuse under CC BY 4.0: credit BenchLeader with a link to this page. CC BY 4.0 (attribution by link to https://www.benchleader.com/stats/ai-race): BenchLeader Index figures as scored today, by release month. No benchmark scores are included.

Embed code

<figure><a href="https://www.benchleader.com/stats/ai-race"><img src="https://www.benchleader.com/stats/ai-race.svg" alt="Claude Opus 5.5 by Anthropic is the best AI model available on the BenchLeader Index (69.2), in the lead since September 2026; the lead has changed hands 9 times since June 2024." width="760"></a><figcaption>Source: <a href="https://www.benchleader.com/stats/ai-race">BenchLeader</a>, CC BY 4.0</figcaption></figure>

Download the table (CSV)

Citation

BenchLeader, “The AI race: the best AI model by month”, data as of 2 Nov 2026, https://www.benchleader.com/stats/ai-race.

Questions people ask

Who is winning the AI race?
Claude Opus 5.5 by Anthropic is the best AI model available on the BenchLeader Index (69.2), in the lead since September 2026; the lead has changed hands 9 times since June 2024. By months in the lead since June 2024: OpenAI (20 months), Anthropic (9 months) and Google (1 month).
How often does the best AI model change?
9 changes at the top in 30 months since June 2024, so a lead has lasted about 3 months on average. The longest was o3's 10 months.
How far behind are open-weights models?
The best open-weights model, Kimi K3 by Moonshot AI, scores 62.1: 7.1 points behind the best closed model, against 4.5 in June 2024.
How is this timeline built?
For each month, the best model on today's BenchLeader Index among those released by the month's end. It is read back on today's scores, not on what was measured at the time, and counts only models with at least 10 independent results, so a model scored on a few boards cannot take the lead. Models BenchLeader no longer tracks are not in it. A lead of a point or two is within the Index's margin of error.

Related: the leaderboard today, new model releases, releases by month, LLM price trends, best open-weights models, AI trends and statistics.