AI model market · 2026-43
DeepSeek V4.1 Flash 1017 rose 2.3 points to 2.3% of OpenRouter tokens
On the 19 October 2026 snapshot, DeepSeek V4.1 Flash 1017 was the biggest riser in OpenRouter token usage, up 2.3 points to 2.3%. The comparison runs since 16 October 2026, not over a full week. DeepSeek V4.1 Flash still led the usage ranking at 20.0%, though its share fell 3.8 points.
Produced automatically from OpenRouter and LMArena data.
- Most-used model
- DeepSeek V4.1 Flash
- Biggest riser
- DeepSeek V4.1 Flash 1017
- Cheapest frontier model
- MiMo-V2.6-Pro
- New models
- 14
20.0%
+2.3
$0.435 / $0.87
What changed
- Step 5 Preview from Stepfun rose 1.9 points to 7.0% of OpenRouter tokens, placing third, while GLM 5.3 Flash from Z.ai rose 0.4 points to 7.5% in second place.
- Gemini 4 Argon led the LMArena text leaderboard with a score of 1,532.3, ahead of Claude Opus 5.5 at 1,515.8 in second and Claude Fable 5.1 at 1,511.6 in third.
- OpenRouter added 14 new models in the past 7 days, including DeepSeek V4.1 Flash 1017 at $0.30 input and $1.20 output per million tokens with a 1.05M-token context.
- OpenRouter listed 22 price changes since 16 October 2026; DeepSeek Pro Latest output rose from $0.7484 to $8.00, while DeepSeek V4.1 Flash output fell from $0.60 to $0.12.
- MiMo-V2.6-Pro was the cheapest frontier model in the LMArena top 15, at $0.435 input and $0.87 output per million tokens, while GLM 5.3 Flash led value by arena score per blended dollar at 17,098.8.
Why it matters
Prices and usage shares are inputs to cost planning for teams that build with or buy AI models. The data shows DeepSeek output prices moving in both directions since 16 October 2026, from a rise to $8.00 on DeepSeek Pro Latest to a fall to $0.12 on DeepSeek V4.1 Flash. A price checked on 16 October 2026 may therefore differ from the price listed on 19 October 2026. Teams that budget by output price may want to check the current listing.
The leaderboard and usage ranking measure different things. Gemini 4 Argon leads LMArena with 1,532.3, while DeepSeek V4.1 Flash holds the largest share of OpenRouter tokens at 20.0%. Buyers comparing models can weigh arena score, price and usage together; MiMo-V2.6-Pro is listed at $0.435 input and $0.87 output per million tokens with a score of 1,490.7. Each measure answers a separate question for buyers.
The numbers
| Rank | Model | Share, 7 days | Change (points) | Input $/1M | Output $/1M |
|---|---|---|---|---|---|
| 1 | DeepSeek V4.1 Flash | 20.0% | -3.8 | $0.0472 | $0.12 |
| 2 | GLM 5.3 Flash | 7.5% | +0.4 | $0.048 | $0.20 |
| 3 | Step 5 Preview | 7.0% | +1.9 | $1.00 | $2.70 |
| 4 | MiMo-V2.6-Flash | 6.1% | -0.6 | $0.14 | $0.28 |
| 5 | Hy4 preview | 4.6% | -0.2 | $0.834 | $2.501 |
| 6 | GPT-6 Luna | 3.8% | -0.3 | $0.10 | $0.50 |
| 7 | Claude Opus 5.5 | 3.8% | +0.9 | $4.00 | $20.00 |
| 8 | GLM 5.3 | 3.2% | +1.0 | $0.079 | $1.40 |
| 10 | DeepSeek V4 Flash 0731 | 2.9% | 0.0 | $0.15 | $0.15 |
| 11 | DeepSeek V4.1 Flash 1017 | 2.3% | +2.3 | $0.30 | $1.20 |
Methodology and sources
Written automatically by a language model from daily snapshots of the OpenRouter model catalogue, the OpenRouter rankings (CC BY 4.0) and the LMArena text leaderboard (CC BY 4.0). Snapshot date 19 October 2026; comparison: since 16 October 2026. The model was given the numbers only and told to add nothing else; numbers and wording are checked by program before publication. Live figures: /ai-models.
Cite this page
Oesnada, "DeepSeek V4.1 Flash 1017 rose 2.3 points to 2.3% of OpenRouter tokens", https://oesnada.com/analysis/ai-model-market/2026-43, accessed 19 October 2026.