What shipped |
● Google · Nano Banana & Imagen · image model · Oct 6 An update to Nano Banana 2 built on Gemini 3.6 Flash that scores 60 Elo higher in Google's blind text-to-image tests at about half the price per image. Compared with Nano Banana 2 (Gemini 3.1 Flash Image) · the model it replaces 1050 ▲ up from 990 for Nano Banana 2 · +60 Elo, margin about 14 · Google's own test, thinking mode Preferred more often than Nano Banana 2 in blind image comparisons · Google text-to-image Elo | $0.0336 / image ▼ down from $0.067 for Nano Banana 2 · 2K image $0.0504 Costs about half as much per image as Nano Banana 2 · Gemini API price per generated image |
|
● Anthropic · Claude Haiku · Oct 7 Costs a tenth of Claude Haiku 4.5 for prompts under 100K tokens while raising its computer-use score from 15.7% to 72.4%, with a 1M-token context window. Compared with Claude Haiku 4.5 · the model it replaces 72.4% ▲ up from 15.7% · +56.7 points Finishes 7 in 10 tasks when operating a computer on its own · OSWorld 2.1 | $0.10 / $0.50 ▼ down from $1 / $5 · prompts under 100K tokens; $0.50 / $2.50 above that Costs developers a tenth of what Claude Haiku 4.5 cost · Price per 1M tokens |
|
● Mistral · Mistral Large · open weights · Oct 6 Mistral's first large reasoning model, natively multimodal with 1 trillion parameters (49 billion active), lifts the Intelligence Index from 9 to 38 but costs almost three times as much as Mistral Large 3. Compared with Mistral Large 3 · the model it replaces 61.7% ✦ first score published — no Mistral Large 3 result to compare Solves about 6 in 10 hard real-world software engineering tasks · DeepSWE 1.1 | $1.36 / $4.18 preview launch price — was $0.50 / $1.50 for Mistral Large 3 Costs developers almost three times as much to run · API price per 1M tokens |
|
Also new this week |
● Xiaomi · MiMo Pro · open weights · released Sep 22 · added this week Xiaomi's new open-weights flagship scores 46 on the Artificial Analysis Intelligence Index, up from 26 for MiMo-V2.5-Pro, at the same API price of $0.435 / $0.87 per 1M tokens. |
● Black Forest Labs · FLUX · image model · released Oct 1 · added this week Generates images natively at 2K and 4K and adds bounding-box layout control to place, re-describe or replace individual elements. |
● Xiaomi · MiMo Flash · open weights · released Sep 22 · added this week Xiaomi's new open-weights Flash model beats its previous Pro model, MiMo-V2.5-Pro, on all 14 tests where Xiaomi's table scores both, while keeping MiMo-V2.5's price of $0.14 / $0.28 per 1M tokens. |
Coming next |
Announced by the maker, not out yet |
● Gemini 4 Argon · Google · announced Sep 30 · expected as soon as possible · open only to selected partners so far “Today, we’re announcing our new frontier model, Gemini 4 Argon, which is rolling out to a set of trusted cyber defenders through our Fairwind Program.” Google |
“Grok 4.8, which is a 2.5T model trained with our new C++ software stack, will finish training this week and start RL” Elon Musk on X |
“all deepseek-v4-pro requests will route to V4.1-Flash at V4.1-Flash rates. This will continue until V4.1-Pro launches.” DeepSeek API docs |
● Qwen 4 · Alibaba · announced Sep 22 “At the conference, Alibaba revealed that its next-generation model, Qwen 4, is currently in training.” Alibaba Cloud |
Close to their usual gap Time since the last release is close to the usual gap between this line’s models. We don’t predict dates. |
● GLM · Z.ai · last: GLM-5.3 59 days since the last release — usual gap is 62 |
● Seed Lite · ByteDance · last: Doubao-Seed-2.0-Lite 240 days since the last release — usual gap is 252 |
133 days since the last release — usual gap is 143 |
|
Next AI Model checks 14 AI makers for new models every morning. Every number comes from the maker or an independent test, with its source. One email a week, only when something ships. |
