Signal SentrySearchUpdated Nov 20, 05:30 UTCAM Drop
Models · scorecard

Mistral Large 4

What the vendor claims, next to what independent boards measured. The two are never mixed in one column.

At a glance

Maker
Mistral
Price per 1M tokens
$0.7 in · $2.1 out
Context window
1.05M
Input → output
text, image → text
Added to OpenRouter
Oct 6, 2026

Source: OpenRouter, as of Nov 20, 06:00 UTC. Model page on OpenRouter

What the vendor claims

Quoted word for word from the vendor's own launch post or model card. Claims, not measurements.

Scores the vendor claims, each with its source and the quoted line
BenchmarkClaimedSource
DeepSWE v1.1The source says: ML4 excels across software engineering, repository understanding, and complex terminal workflows, scoring 61.7% on DeepSWE v1.1, 59.4% on SWE-Atlas-QnA, and 28.3% on Terminal-Bench 4.61.7%Mistral · Oct 6From Drop #010 · Wed, Oct 7, 2026 · Evening
SWE-Atlas-QnAThe source says: ML4 excels across software engineering, repository understanding, and complex terminal workflows, scoring 61.7% on DeepSWE v1.1, 59.4% on SWE-Atlas-QnA, and 28.3% on Terminal-Bench 4.59.4%Mistral · Oct 6From Drop #010 · Wed, Oct 7, 2026 · Evening
Terminal-Bench 4The source says: ML4 excels across software engineering, repository understanding, and complex terminal workflows, scoring 61.7% on DeepSWE v1.1, 59.4% on SWE-Atlas-QnA, and 28.3% on Terminal-Bench 4.28.3%Mistral · Oct 6From Drop #010 · Wed, Oct 7, 2026 · Evening
CybenchThe source says: It also solves 93% of the challenges in Cybench, a set of 40 exercises drawn from security competitions, one of the highest scores reported for an open-weight model.93%Mistral · Oct 6From Drop #010 · Wed, Oct 7, 2026 · Evening
AutomationBenchThe source says: On AutomationBench — 657 business workflows across apps like Gmail, Google Sheets, Slack, and Salesforce — it scores 59.9%, ahead of Kimi K3, MiMo-V2.6-Pro, and DeepSeek V4 Pro.59.9%Mistral · Oct 6From Drop #010 · Wed, Oct 7, 2026 · Evening
Lakera B3 AI Security BenchmarkThe source says: On Lakera’s public B3 AI Security Benchmark, ML4 resists 93.3% of attacks – we see no higher scores among competitors.93.3%Mistral · Oct 6From Drop #010 · Wed, Oct 7, 2026 · Evening

Independent results

Measured by independent sources (Epoch AI, LMArena, OpenRouter's benchmark runs, τ²-bench, Vectara), not by the vendor. "Listed as" is the source's own name for the model, so you can check our match.

No independent results yet. They appear here as Epoch AI and LMArena add this model.

Price history

OpenRouter's list price per 1M tokens in each of our snapshots.

  • Output now $2.1
  • Input now $0.7

Changed 2 times across our 90 snapshots.

Show the numbers
This model's OpenRouter price in each of our snapshots
SnapshotInput $/1MOutput $/1M
Nov 20, 06:04 UTC$0.7$2.1
Nov 19, 16:04 UTC$0.7$2.1
Nov 19, 04:04 UTC$0.7$2.1
Nov 18, 18:04 UTC$0.7$2.1
Nov 18, 04:03 UTC$0.7$2.1
Nov 17, 16:01 UTC$0.7$2.1
Nov 17, 04:04 UTC$0.7$2.1
Nov 16, 19:54 UTC$0.7$2.1
Nov 16, 04:05 UTC$0.7$2.1
Nov 15, 16:04 UTC$0.7$2.1
Nov 15, 04:03 UTC$0.7$2.1
Nov 14, 16:04 UTC$0.7$2.1
Nov 14, 04:05 UTC$0.7$2.1
Nov 13, 16:05 UTC$0.7$2.1
Nov 13, 04:04 UTC$0.7$2.1
Nov 12, 16:05 UTC$0.7$2.1
Nov 12, 04:04 UTC$0.7$2.1
Nov 11, 16:01 UTC$0.7$2.1
Nov 11, 04:04 UTC$0.7$2.1
Nov 10, 16:04 UTC$0.7$2.1
Nov 10, 04:05 UTC$0.7$2.1
Nov 9, 16:04 UTC$0.7$2.1
Nov 9, 04:04 UTC$0.7$2.1
Nov 8, 16:04 UTC$0.7$2.1
Nov 8, 04:03 UTC$0.7$2.1
Nov 7, 16:04 UTC$0.7$2.1
Nov 7, 04:04 UTC$0.7$2.1
Nov 6, 16:04 UTC$0.7$2.1
Nov 6, 04:04 UTC$0.7$2.1
Nov 5, 16:05 UTC$0.7$2.1
Nov 5, 10:05 UTC$0.7$2.1
Nov 4, 16:02 UTC$0.7$2.1
Nov 4, 04:04 UTC$0.7$2.1
Nov 3, 16:04 UTC$0.7$2.1
Nov 3, 13:26 UTC$0.7$2.1
Nov 2, 16:01 UTC$0.7$2.1
Nov 2, 04:04 UTC$0.7$2.1
Nov 1, 16:03 UTC$0.7$2.1
Nov 1, 04:04 UTC$0.7$2.1
Oct 31, 16:04 UTC$0.7$2.1
Oct 31, 04:04 UTC$0.7$2.1
Oct 30, 16:03 UTC$0.7$2.1
Oct 30, 04:04 UTC$0.7$2.1
Oct 29, 16:03 UTC$0.7$2.1
Oct 29, 04:04 UTC$1.36$4.18
Oct 28, 16:04 UTC$1.36$4.18
Oct 28, 09:22 UTC$1.36$4.18
Oct 27, 16:03 UTC$1.36$4.18
Oct 27, 04:03 UTC$1.36$4.18
Oct 26, 16:04 UTC$1.36$4.18
Oct 26, 04:03 UTC$1.36$4.18
Oct 25, 16:04 UTC$1.36$4.18
Oct 25, 06:03 UTC$1.36$4.18
Oct 24, 15:04 UTC$1.36$4.18
Oct 24, 03:03 UTC$1.36$4.18
Oct 23, 15:03 UTC$1.36$4.18
Oct 23, 03:04 UTC$1.36$4.18
Oct 22, 15:03 UTC$1.36$4.18
Oct 22, 05:04 UTC$1.36$4.18
Oct 21, 15:03 UTC$1.36$4.18
Oct 21, 03:02 UTC$1.36$4.18
Oct 20, 15:04 UTC$0.68$2.09
Oct 20, 03:03 UTC$0.68$2.09
Oct 19, 15:04 UTC$0.68$2.09
Oct 19, 10:50 UTC$0.68$2.09
Oct 18, 15:03 UTC$0.68$2.09
Oct 18, 03:05 UTC$0.68$2.09
Oct 17, 15:03 UTC$0.68$2.09
Oct 17, 03:03 UTC$0.68$2.09
Oct 16, 15:02 UTC$0.68$2.09
Oct 16, 03:03 UTC$0.68$2.09
Oct 15, 17:02 UTC$0.68$2.09
Oct 15, 03:02 UTC$0.68$2.09
Oct 14, 15:03 UTC$0.68$2.09
Oct 14, 03:02 UTC$0.68$2.09
Oct 13, 15:03 UTC$0.68$2.09
Oct 13, 05:03 UTC$0.68$2.09
Oct 12, 15:03 UTC$0.68$2.09
Oct 12, 03:04 UTC$0.68$2.09
Oct 11, 15:02 UTC$0.68$2.09
Oct 11, 03:03 UTC$0.68$2.09
Oct 10, 15:01 UTC$0.68$2.09
Oct 10, 03:04 UTC$0.68$2.09
Oct 9, 15:05 UTC$0.68$2.09
Oct 9, 03:04 UTC$0.68$2.09
Oct 8, 15:03 UTC$0.68$2.09
Oct 8, 03:05 UTC$0.68$2.09
Oct 7, 15:03 UTC$0.68$2.09
Oct 7, 03:03 UTC$0.68$2.09
Oct 6, 15:03 UTC$0.68$2.09

Source: OpenRouter, as of Nov 20, 06:04 UTC

In the Drops

ModelsFollowing since Oct 6, 2026in 3 Drops

Mistral says Large 4 was trained from scratch, with weights due by month's end

Mistral says Large 4 is a 1 trillion-parameter model with 49 billion active, trained from scratch on 3,800 Nvidia Grace Blackwell GPUs in its own European datacenters. A preview API is live now.

Why it matters: Ars, republishing Wired, says Mistral pitches it as insurance against losing access to closed models. OpenRouter lists it at $0.68 in and $2.09 out per 1M tokens.

From Drop #010 · Wed, Oct 7, 2026 · Evening