Mistral Small 3.2 24B Instruct
mistral-small-3-2-24b-instructmistral-small-3.2-24b-instruct- - 🏢 Built by Mistral AI, a Paris-based open-weight LLM company.
- - 🆕 Incremental update to Mistral Small 3.1, same 24B architecture.
- - 🎯 Tuned for better instruction following and fewer repetition errors.
- - 🔧 Improved function/tool calling and structured-output reliability.
- - 👁️ Multimodal: handles both image and text inputs.
- - 📏 Long-context support, with this deployment exposing a large window.
- - 🔒 Released under the permissive Apache 2.0 license.
- - ⚡ Optimized for efficiency, balancing speed and capability.
Mistral AI is a French artificial intelligence company headquartered in Paris, founded in 2023. The company focuses on developing large language models offered under both open-weight and proprietary licenses. Mistral AI has quickly risen to prominence in the AI landscape,…
Explore 1 more model by Mistral AI →Mistral Small 3.2 24B Instruct is a 24-billion-parameter model from Mistral AI, a French lab known for open-weight releases under permissive licenses. According to Mistral's own documentation, it is an update of Mistral-Small-3.1-24B-Instruct, keeping the same 24B architecture while refining behavior rather than scaling up. It is distributed under Apache 2.0 and supports both image and text inputs, function calling, and structured outputs.
Compared with its 3.1 predecessor, Mistral's documentation describes version 3.2 as focusing on enhanced instruction following, reduced repetition and "infinite generation" errors, and more robust function calling and structured output. The lab frames this as a targeted refinement of the existing model rather than a broad capability leap, with most other categories expected to match or slightly improve on Small 3.1.
On this Venice deployment, the model runs in fp8 quantization with a large context window and adds web-search capability, alongside its native function calling.
Within the same family, the newer Mistral Small 4 is a larger, later release, marking a different design point from this efficiency-oriented 24B model. For users wanting a compact, openly licensed model that balances speed with multimodal and tool-use capability, Mistral Small 3.2 remains a practical general-purpose option.
This About section is AI-generated from public sources via VeniceStats + Venice inference, with no human editing. It may contain inaccuracies.
Daily usage · Mistral Small 3.2 24B Instruct · Organic sellers
Top sellers · Mistral Small 3.2 24B Instruct · Organic
- Venice.ai Proxy96.3%
- Fire Ant 🔥🐜2.3%
- JOI SEED1.3%
- Sapiens0.0%
- Meridian AI0.0%
| Seller | Health | Rep↓ | Sales | Tokens | Routing | Input $/M | Cached $/M | Output $/M | Save | 1st token | tok/s |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Super Seeder 0xd19f…41f3 | healthy | 89.22 | $0.00 | 112.67k | #1 | $0.00 | $0.00 | $0.00 | −100% | 0.6 s | — |
| Sapiens 0x1fae…8318 | healthy | 8.20 | $0.00 | 1.92k | gated | $0.0717 | $0.0717not advertised | $0.1913 | −28% | — | — |
| ➤Bullet Ant 🐜 0xe924…8936 | — | 0.00 | — | — | gated | $0.50 | $0.50not advertised | $1.50 | — | — | — |
| surplus-hive 0x6145…5c58 | — | 0.00 | — | — | gated | $0.2279 | $0.2279 | $0.6076 | — | — | — |
| Fire Ant 🔥🐜 0xbe05…bc5d | — | 0.00 | $0.01 | 142.24k | gated | $0.0617 | $0.0617not advertised | $0.1813 | −38% | — | — |
| Joi Gateway 0x8a90…7007 | — | 0.00 | — | — | gated | $0.04 | $0.04 | $0.1068 | −60% | — | — |
| antseed-neon-puma-944e 0x6650…944e | — | 0.00 | — | — | gated | $0.032 | $0.032not advertised | $0.0853 | −68% | — | — |
| Meridian AI 0x8c8c…06f5 | — | 0.00 | $0.00 | 280.00 | gated | $0.0091 | $0.0091not advertised | $0.0243 | −91% | — | — |
| Leftermute 0x388b…5389 | — | 0.00 | — | — | gated | $0.0085 | $0.0085not advertised | $0.0227 | −92% | — | — |
| Venice.ai Proxy 0x1f22…18c9 | failingupstream error · 3.7 h · stale | 86.84 | $0.47 | 9.22M | gated | $0.0516 | $0.0516not advertised | $0.1375 | −48% | — | — |
"Best price" and the seller table are live Antseed catalog data (advertised $/1M tokens — or $ per generated image for unit-billed image models — not settled amounts). Reputation = buyer trust score (0-100, the Antseed SDK's own formula). "Routing" = the SDK's default buyer routing (what the VPR desktop app ships with): a trust ≥ 60 gate on the effective reputation, then cheapest-first among routable sellers; live failover state (per-peer cooldowns) is buyer-side runtime and not included. Model knowledge (TLDR, provider, About) via the VeniceStats enrichment layer. Advertised catalog, not the model used in any specific purchase. "Usage on Antseed" counts only settlements whose buyers share the per-model split on-chain (metadata v2/v3, opt-in), so every usage figure is a lower bound.