Devstral Small 24B · 11 GPUs measured first-party · llama.cpp Q4_K_M · Updated October 2026

What GPU Do You Need for Devstral Small 24B?

Devstral Small 24B is Mistral's agent-tuned coder, built with All Hands AI specifically for software-engineering agent workflows: navigating repos, editing multiple files, running loops until tests pass. Measured on 11 GPUs (llama.cpp, Q4_K_M): 121 tok/s on the B300, ~15GB peak VRAM.

Benchmarked weights: bartowski/mistralai_Devstral-Small-2507-GGUF

Fastest we measured

NVIDIA B300

121.2 tok/s on Devstral Small 24B, the ceiling. Measured on our bench. 288GB of VRAM, $40,000 at launch.

Pros

• 121.2 tok/s on Devstral Small 24B
• 288GB, clears the Devstral Small 24B floor
• Rentable by the hour rather than bought

Cons

• 1400W board rating
• Datacenter or workstation hardware, not a retail purchase

Cheapest card that runs it

NVIDIA RTX PRO 6000 Blackwell Workstation Edition

93.52 tok/s on Devstral Small 24B, lowest launch price that still fits. Measured on our bench. 96GB of VRAM, $8,565 at launch.

Pros

• 93.52 tok/s on Devstral Small 24B
• 96GB, clears the Devstral Small 24B floor
• Rentable by the hour rather than bought

Cons

• 600W board rating
• Datacenter or workstation hardware, not a retail purchase

The agent-loop specialist. Most coder models are trained to produce code; Devstral is trained to *behave*: to operate inside an agent harness, chain tool calls, and keep multi-file state straight. That's a different skill, and it's why this model exists in our lineup as the counterpart to Codestral: completion is Codestral's lane, agency is Devstral's. If you're building toward local autonomous coding, an agent that takes an issue and produces a diff, this is the mid-size model designed for exactly that loop.

What the bench says. 121 tok/s peak, and an H200 sweet spot of 109 tok/s at 138W (0.79 tok/W, the class efficiency lead). Agent loops multiply token volume, every step re-reads context and emits edits, so sustained throughput and prompt speed (3,842 t/s on the B200) matter more than for chat use. The ~15GB floor mirrors the Dolphin Mistral 24Bs: 16GB cards fit it snugly, 24GB is the comfortable recommendation once real contexts pile up.

How it compares. H100 80GB HBM3: Devstral Small 24B 107.8 tok/s, Dolphin 3.0 R1 Mistral 24B 107.8, Dolphin Mistral 24B Venice 109.1, Dolphin-Mistral-24B-Venice-Edition 105.6, Mistral Small 24B 105.6. 2 of 4 beat Devstral Small 24B here.

Cost on a rented GPU. 1M generated tokens of Devstral Small 24B: $2.10 on a A100 40GB SXM4 ($0.47/hr, 4.5 hours), $15.91 on a B300 ($6.94/hr, 2.3 hours, 7.6x the cost).

Speed tiers for Devstral Small 24B. 30+ tok/s: 9 (B300, B200, H200); 10-30 tok/s: 2 (L4, T4). 30 tok/s is roughly where replies outpace reading.

Reading your prompt. Before Devstral Small 24B writes anything it reads the input: 4840.7 tok/s on the RTX PRO 6000 Blackwell Workstation Edition (0.8s for a 4,000-token prompt), 421.0 on the T4 (9.5s). Long documents and big code files feel this number more than the generation speed.

VRAM for Devstral Small 24B. Measured peak 13.9GB, so 16GB is the smallest common card size; smallest card it ran on: T4 (16GB). With long context: Q4_K_M 16GB (tested), Q2_K 10GB, Q3_K_M 13GB, Q5_K_M 19GB, Q6_K 22GB.

Power on Devstral Small 24B. Most efficient: H100 80GB HBM3, 140W, 0.36 kWh per 1M generated tokens. Hungriest: B200, 374W, 0.91 kWh. At $0.15/kWh: $0.054 per 1M generated tokens.

Our verdict

Devstral Small 24B: 121 tok/s peak, ~15GB floor, the model in our database purpose-built for coding agents rather than coding chat. Run it in the loop, keep Codestral in the editor, and a single 24GB card hosts the complete division of labor.

FAQ

What makes Devstral different from other coder models?

Training target. It was tuned (by Mistral with All Hands AI) for agentic software engineering, repo navigation, multi-file edits, tool-call chains, not just emitting code snippets. In agent harnesses that behavioral training is the difference between finishing tasks and wandering.

What GPU does it need?

~15GB measured peak at Q4_K_M: a 16GB card fits with little headroom, 24GB is comfortable. Speed peaked at 121 tok/s (B300), with the H200's 109 tok/s at 138W the efficiency winner.

Why does throughput matter extra for agent use?

Agent loops are token-hungry: every iteration re-ingests context and generates edits, multiplying volume 10-50× over a chat exchange. Prompt speed and sustained tok/s directly set how long an autonomous task takes.

Devstral or Qwen3 Coder 30B-A3B for agents?

The Qwen MoE is faster (318 vs 121 tok/s peak) but needs ~20GB; Devstral's agent-specific tuning and 16GB viability are its cards. On a 24GB rig, test both in your harness, behavioral fit tends to decide it.

How does it pair with Codestral?

Perfectly. That's the intended stack: Codestral 22B answers your editor's fill-in-middle requests; Devstral drives the agent loop. Both are Mistral-family, both fit 16GB-class cards, and together they cover the two halves of real coding work.