Mistral AI
Timestral-123M is a proprietary Mistral AI specialized model.
Live access is not confirmed for this model. No current purchase price is advertised.
No current subscription pricing is tracked for this model.
Confirm this specific model, usage limits, and billing terms with the provider. A subscription does not automatically include API credits.
No login needed to compare. Prices are in USD; provider charges are separate from AI Market Cap plans. Context length, caching, tools, taxes, and regional terms can change the final cost. Open weights do not mean free hosting.
35.1
Quality Score
---
Arena ELO
123M
Parameters
---
Context
This measures the amount of verifiable public evidence we have, not how capable the model is. A missing field means it has not been verified yet, not that its value is zero.
12 of 22 public signals
Sign in to join the discussion
8.8K
Downloads
47
Likes
Oct 2026
Released
3/5 signals
2/4 signals
2/5 signals
3/4 signals
2/4 signals
Gaps we are still tracking
Launches
2
Benchmarks
1
Research
3
General
5
Recent launch, pricing, benchmark, and API signals linked to this model or its provider.
Today, we’re introducing the Responses API in public preview, made for building agents on Mistral. Each response returns the model’s answer, its reasoning and its tool calls as separate items, so your app can handle every step more easily than with Chat Completions. It works https://t.co/5zgjZgqugd
View sourceToday, we release the weights of a new and improved Le Chonk, advancing the open frontier for agentic coding and cybersecurity. 💻 Built to build: Le Chonk ranks #1 on @cognition’s FrontierCode among open models, and sets a new benchmark for Pareto efficiency on https://t.co/R9wQ6WTPnl
View sourceWe introduce Timestral, a family of four time series foundation models for probabilistic forecasting, with 5M, 28M, 123M, and 209M parameters. Timestral sets a new state of the art on GIFT-Eval, TIME, and fev-bench, where Timestral-209M and Timestral-123M take the top two places on every metric. We achieve this with three ingredients: Decoupled Horizon Forecasting (DHF), which separates prediction tokens from the causally encoded context, so that longer horizons are trained without masking context and each observed value is both a forecast target and context for later forecasts; a covariate framework that natively supports known-future covariates within a causal architecture; and a two-stage pipeline that pretrains exclusively on a diverse, large-scale synthetic corpus before post-training on a mix of real and synthetic data with extended context. This report describes the model design and its benchmark results. Weights for all four models are available on Hugging Face and inference code on GitHub.
View sourceWhat does real-time transcription look like in a live setting? 🎙 We used Voxtral Transcribe 3 at AI Engineer Paris to capture every main-stage talk in real time. We boosted accuracy with context biasing, feeding the model speaker names and technical terms from the conference https://t.co/L2mxmjrlav
What does real-time transcription look like in a live setting? We used Voxtral Transcribe 3 at AI Engineer Paris to capture every main-stage talk in real time. We boosted accuracy with context biasing, feeding the model speaker names and technical terms from the conference https://t.co/UKbQD6DJPk
What does real-time transcription look like in a live setting? 🎙We used Voxtral 3 Transcribe at AI Engineer Paris to capture every main-stage talk in real time. We boosted accuracy with context biasing, feeding the model speaker names and technical terms from the conference https://t.co/Fgf1Py4Z6I

Today, we release the weights of a new and improved Le Chonk, advancing the open frontier for agentic coding and cybersecurity. 💻 Built to build: Le Chonk ranks #1 on @cognition’s FrontierCode among open models, and sets a new benchmark for Pareto efficiency on https://t.co/R9wQ6WTPnl

Today, we’re introducing the Responses API in public preview, made for building agents on Mistral. Each response returns the model’s answer, its reasoning and its tool calls as separate items, so your app can handle every step more easily than with Chat Completions. It works https://t.co/5zgjZgqugd

Today we’re announcing a new collaboration with @nvidia to bring AI to industrial engineering: - faster simulation - open benchmarks and supporting datasets for physics AI - evaluations for agentic AI in CAE and EDA More here: https://t.co/HXFVBqJSP1 https://t.co/OTb24yjOUQ

What can large-scale AI models bring to the growing complexity of particle physics? @CERN and Mistral join forces to explore this question. The ambition: a single Large Physics Model, trained on billions of simulated collisions. https://t.co/nrdCfcBI3k
We introduce Timestral, a family of four time series foundation models for probabilistic forecasting, with 5M, 28M, 123M, and 209M parameters. Timestral sets a new state of the art on GIFT-Eval, TIME, and fev-bench, where Timestral-209M and Timestral-123M take the top two places on every metric. We achieve this with three ingredients: Decoupled Horizon Forecasting (DHF), which separates prediction tokens from the causally encoded context, so that longer horizons are trained without masking context and each observed value is both a forecast target and context for later forecasts; a covariate framework that natively supports known-future covariates within a causal architecture; and a two-stage pipeline that pretrains exclusively on a diverse, large-scale synthetic corpus before post-training on a mix of real and synthetic data with extended context. This report describes the model design and its benchmark results. Weights for all four models are available on Hugging Face and inference code on GitHub.
Large Language Model inference over long contexts is increasingly bottlenecked by the cost of attending to a growing Key-Value (KV) cache. Sparse attention reduces this cost by accessing only a fraction of the cache, but existing methods face a trade-off between deterministic retrieval, which captures high-scoring tokens but discards residual attention mass, and sampling, which improves coverage at the cost of higher variance. Hybrid methods combine top-k retrieval with uniform sampling, yet still exhibit a substantial gap from dense attention at aggressive sparsity. We introduce Stratified Sparse Attention (SSA), a unified framework combining deterministic retrieval with score-guided sampling. SSA deterministically attends to high-scoring tokens while sampling the residual using proxy attention scores and correcting for their inclusion probabilities, thereby recovering attention mass beyond the retrieval set. SSA encompasses top-k, top-p, and hybrid top-k+sampling as special cases. Across RULER, LongBench, LOFT, and AIME, and five models ranging from 3B to 27B parameters, SSA consistently improves over the baselines at matched sparsity. On RULER-HARD with Ministral-3-8B, SSA improves accuracy over the strongest practical baseline from 81.8\% to 84.2\% at 32K and from 63.8\% to 69.3\% at 128K under 50times attention sparsity, recovering 94\% and 85\% of dense-attention performance, respectively. With optimized CUDA kernels, SSA achieves up to 2.96times decode speedup over FlashInfer at 1M context, demonstrating that its accuracy gains translate into practical long-context inference acceleration. We open-source our code at https://github.com/skylight-org/SSA-Paper-Code.