artificialwatch

The best AI model right now is Claude Opus 5.5

Anthropic's model leads ArtificialWatch's combined ranking of 6 public leaderboards and is first on 5 of the 6 on its own. Next: Claude Fable 5.1, then GPT-6 Astra.

boards read Sep 28 · 31 models ranked · computed from public scores, not our opinion · updated daily

Know the minute a new #1 ships →

#1 on each board

each board's own top score · read Sep 28

Claude Opus 5.5 is first on 5 of 6. On LiveBench, Claude Fable 5.1 leads at 83.4; Claude Opus 5.5 scores 83.2.

top 10, combined

bench index: median percentile across the boards that score each model
  1. 01Claude Opus 5.5Anthropic · 6 boards100$4 / $20
  2. 02Claude Fable 5.1Anthropic · 6 boards96.3$10 / $50
  3. 03GPT-6 AstraOpenAI · 6 boards91.5$10 / $50
  4. 04Claude Opus 5Anthropic · 6 boards85.9$5 / $25
  5. 05Claude Fable 5Anthropic · 6 boards85.1$10 / $50
  6. 06Muse Spark 1.3Meta · 6 boards81.4$1.25 / $4.25
  7. 07GPT-6 SolOpenAI · 5 boards75.9$2 / $10
  8. 08GPT-5.6OpenAI · 6 boards70.2$4 / $20
  9. 09Grok 4.7xAI · 4 boards69.4$2 / $6
  10. 10Grok 4.6xAI · 5 boards60$2 / $6

Price: per 1M input / output tokens on the maker's own API, as listed September 27, 2026. Every board's number for all 31 models: the full bench index. Every price: API prices.

most model per dollar

25 models with a published price
  1. Muse Spark 1.3$2 blended · index 81.481% of the #1's index for 25% of its price
  2. Gemini 3.8 Flash$1.50 blended · index 5050% of the #1's index for 19% of its price
  3. GPT-6 Luna$0.20 blended · index 40.941% of the #1's index for 3% of its price

The line joins every model that outscores all cheaper ones: along it, no model with a published price scores higher for less. Blended price is 3 input tokens to 1 output, per million. Models without a list price are left off.

what traders pay

polymarket · which company has the best AI model
  1. END SEP$4.1M traded · rules ↗
    99%+Anthropic

    everyone else under 1%

It settles on the arena.ai Text Arena ↗, as its rules name it, at noon ET on the date. Later dates and every AI release market: AI model odds.

sweeping every 60 seconds

The #1 changes when a new model ships. Know the minute it answers.

A free Chrome push the moment a new model answers on a public API, an email about 15 minutes later. Texts or a phone call on paid plans.

  • Free forever
  • No card
  • Unsubscribe in one click

Questions

What is the best AI model right now?

Claude Opus 5.5 from Anthropic. It leads ArtificialWatch's combined ranking of 6 public leaderboards and is first on 5 of the 6 on its own. The ranking takes the median of each model's percentile across the boards that score it, so no single board decides it. Next are Claude Fable 5.1 and GPT-6 Astra.

Which AI model is the best for the money?

By the makers' own API prices, blended at 3 input tokens to 1 output: GPT-6 Luna ($0.20 per million, index 40.9), Gemini 3.8 Flash ($1.50 per million, index 50) and Muse Spark 1.3 ($2 per million, index 81.4), then Claude Opus 5.5 at $8. No model with a published price scores higher than any of them for less. Muse Spark 1.3 reaches 81% of Claude Opus 5.5's index at 25% of its price.

Which company has the best AI model according to Polymarket?

On Polymarket's "Which company has the best AI model end of September" market, Anthropic is at 99%+. That market settles on the arena.ai Text Arena leaderboard, as its rules name it; these are live prices, not a benchmark result.

Which leaderboards is this ranking built from?

LMArena, LiveBench, SimpleBench, LLM Stats, Vals AI and Artificial Analysis. Their #1 models today: LMArena, Claude Opus 5.5; LiveBench, Claude Fable 5.1; SimpleBench, Claude Opus 5.5; LLM Stats, Claude Opus 5.5; Vals AI, Claude Opus 5.5; Artificial Analysis, Claude Opus 5.5.