As of 5 November 2026, no option is proven ahead. 3 options of 21 cannot be separated with the evidence in hand: Gemini 4 Carbon (High), Gemini 4 Argon (High) and Claude Opus 5.5 (High).
- Options
- 21
- Measured
- 21
- Shortlist
- 3
- Plotted
- 16
- Priced
- 16
The job
Use a hosted AI model through its vendor's own API for general chat and writing, and judge it by how often people prefer its answers in blind comparisons on Arena, against what a stated monthly volume of tokens costs.
Ranking
21 options measured, shortlisted rows shaded
| # | Option | Score 0 to 100 | Cost per month | Interval | Price page |
|---|---|---|---|---|---|
| 1 | Gemini 4 Carbon (High)googleShortlist | 80.0 | Not priced | ci9576.9 to 83.1 | |
| 2 | Gemini 4 Argon (High)googleFrontierShortlist | 76.2 | $40.00 | ci9574.1 to 78.3 | |
| 3 | Claude Opus 5.5 (High)anthropicShortlist | 71.9 | $80.00 | ci9569.7 to 74.1 | |
| 4 | Claude Fable 5.1 (Max)anthropic | 71.3 | $200.00 | ci9569.3 to 73.2 | |
| 5 | Claude Opus 5 (Max)anthropic | 69.3 | $100.00 | ci9567.7 to 70.9 | |
| 6 | Claude Opus 5 (High)anthropic | 68.0 | $100.00 | ci9566.8 to 69.2 | |
| 7 | Claude Opus 4.6 (High)anthropic | 67.2 | $100.00 | ci9566.2 to 68.3 | |
| 8 | Muse Wonder (Max)meta | 66.4 | Not priced | ci9564.0 to 68.8 | |
| 9 | Gemini 3.8 Flash (High)googleFrontier | 66.2 | $15.00 | ci9564.7 to 67.6 | |
| 10 | Claude Opus 4.6anthropic | 65.4 | $100.00 | ci9564.4 to 66.5 | |
| 11 | Claude Fable 5 (High)anthropic | 63.9 | $200.00 | ci9562.5 to 65.2 | |
| 12 | MiMo-V2.6-Proxiaomi | 63.8 | Not priced | ci9560.7 to 66.9 | |
| 13 | Muse Spark 1.3 (Max)meta | 63.5 | Not priced | ci9561.8 to 65.3 | |
| 14 | Claude Opus 4.7 (High)anthropic | 63.2 | $100.00 | ci9562.0 to 64.5 | |
| 15 | Gemini 3.7 Flash (High)google | 63.0 | Not priced | ci9561.5 to 64.6 | |
| 16 | Gemini 3.6 Flash (High)google | 59.9 | $15.00 | ci9558.5 to 61.4 | |
| 17 | Gemini 3 FlashgoogleFrontier | 55.2 | $11.00 | ci9553.8 to 56.7 | |
| 18 | Grok 4.20 Beta (Reasoning, 0309)xai | 50.3 | $17.50 | ci9549.1 to 51.5 | |
| 19 | Grok 4.20 Multi-Agent Beta (0309)xai | 50.2 | $17.50 | ci9549.0 to 51.4 | |
| 20 | Gemini 3.5 Flash-LitegoogleFrontier | 45.2 | $8.00 | ci9543.7 to 46.6 | |
| 21 | GPT-5.6 Luna (xHigh)openaiFrontier | 43.7 | $4.40 | ci9542.3 to 45.1 |
Order is by score, highest first. Intervals overlap, so neighbouring ranks do not differ with certainty. Cost is in USD per month; bars start at zero. Interval: ci95, built from 95% confidence intervals given by the source.
Price against performance
Score against cost of the monthly token workload, log scale
- Frontier: no other option is both cheaper and scores higher
- Other option
- Ring: on the shortlist, not separable with certainty
- Line: score interval, ci95 (built from 95% confidence intervals given by the source)
5 options not plotted: a coordinate is unknown
- Gemini 3.7 Flash (High): Cost unknown
- Gemini 4 Carbon (High): Cost unknown
- MiMo-V2.6-Pro: Cost unknown
- Muse Spark 1.3 (Max): Cost unknown
- Muse Wonder (Max): Cost unknown
Cost for this job
Date read and source for all 21 options
Cost of the monthly token workload, in USD per month, cheapest first within each group.
Workload and assumptions
Cost of 10 million input tokens and 2 million output tokens a month at the vendor's list price per million tokens, read from the vendor's own API price page. Reasoning tokens, prompt caching, batch discounts and long-prompt surcharges are not counted, so a model that thinks longer costs more than shown. The effort in an Arena model name, such as High, does not change the price used here.
| Option | Cost | Price read | Source | Notes |
|---|---|---|---|---|
| Cost 16 | ||||
GPT-5.6 Luna (xHigh)openai | $4.40 | 0 days old | ||
Gemini 3.5 Flash-Litegoogle | $8.00 | 0 days old | ||
Gemini 3 Flashgoogle | $11.00 | 0 days old | ||
Gemini 3.6 Flash (High)google | $15.00 | 0 days old | ||
Gemini 3.8 Flash (High)google | $15.00 | 0 days old | ||
Grok 4.20 Beta (Reasoning, 0309)xai | $17.50 | 0 days old | ||
Grok 4.20 Multi-Agent Beta (0309)xai | $17.50 | 0 days old | ||
Gemini 4 Argon (High)google | $40.00 | 0 days old | ||
Claude Opus 5.5 (High)anthropic | $80.00 | 0 days old | ||
Claude Opus 4.6anthropic | $100.00 | 0 days old | ||
Claude Opus 4.6 (High)anthropic | $100.00 | 0 days old | ||
Claude Opus 4.7 (High)anthropic | $100.00 | 0 days old | ||
Claude Opus 5 (High)anthropic | $100.00 | 0 days old | ||
Claude Opus 5 (Max)anthropic | $100.00 | 0 days old | ||
Claude Fable 5.1 (Max)anthropic | $200.00 | 0 days old | ||
Claude Fable 5 (High)anthropic | $200.00 | 0 days old | ||
| No cost available 5 | ||||
Gemini 3.7 Flash (High)google | Unknown | Not read | No source | 1 note
|
Gemini 4 Carbon (High)google | Unknown | Not read | No source | 1 note
|
MiMo-V2.6-Proxiaomi | Unknown | Not read | No source | 1 note
|
Muse Spark 1.3 (Max)meta | Unknown | Not read | No source | 1 note
|
Muse Wonder (Max)meta | Unknown | Not read | No source | 1 note
|
What evidence would settle it
The three conditions and what is missing
- Claude Opus 5.5 (High): another independent origin or repeats (interval 4.4 points wide), metric Arena text rating (Elo-style, overall category).
- Gemini 4 Argon (High): another independent origin or repeats (interval 4.2 points wide), metric Arena text rating (Elo-style, overall category).
- Gemini 4 Carbon (High): another independent origin or repeats (interval 6.2 points wide), metric Arena text rating (Elo-style, overall category).
The three conditions for naming a best option
- The leader's lower bound must exceed every other option's upper bound by at least 3 points. Not met: the closest gap is -1.4 points (Gemini 4 Carbon (High) against Gemini 4 Argon (High)).
- Every critical metric must come from two independent origins, or from one auditable test of our own. Not met: 1 origin and no auditable own test for Gemini 4 Carbon (High).
- The price must be verified within 30 days. Not met.
Sources and limits
5 sources, the method notes and this board as JSON
Scores on this board come from Arena Leaderboard Dataset (lmarena-ai/leaderboard-dataset). We rescale them to a 0 to 100 scale against fixed anchors (Arena text rating (Elo-style, overall category): 1300 counts as 0 and 1600 counts as 100), so a new option never moves another option's score. The licence and our changes are stated in the data credit at the foot of this page.
- Arena Leaderboard Dataset (lmarena-ai/leaderboard-dataset) free to use with credit; checked
- Claude API pricing prices only, as facts with a link; checked
- Gemini API pricing prices only, as facts with a link; checked
- OpenAI API pricing prices only, as facts with a link; checked
- xAI API models and pricing prices only, as facts with a link; checked
Notes on this board. Only one origin (Arena) measures quality, so no board entry can be named Best until a second independent origin or an own test exists. Arena's rating comes from blind votes by its own users on their own prompts; it is not a test of your job. Prices are read from each vendor's own API price page and are list prices for the stated workload; a model whose price could not be read on the vendor's page is listed with unknown cost and is not plotted. Selection: the 15 models with the highest Arena rank in the overall category, plus every model ranked 16 to 100 whose price was read on a vendor page and whose output price is at most $5 per million tokens.
Scores are on a 0 to 100 scale against editorial anchors that are not yet validated. Prices are for the workload above and change without notice. Evidence older than 60 days is left out, and prices older than 30 days are flagged.
This board as JSON, with every input id, anchor and weight.