Skip to content
llmwaves

V3.1 Terminus

DeepSeek · released Sep 22, 2025

Open weightsReasoningTool useStructured output

DeepSeek-V3.1 Terminus is an update to [DeepSeek V3.1](/deepseek/deepseek-chat-v3.1) that maintains the model's original capabilities while addressing issues reported by users, including language consistency and agent capabilities, further optimizing the model's...

Specification

Context window
164K
Max output
33K
Knowledge cutoff
Mar 31, 2025
Parameters
684.5B
Licence
mit
Serving providers
5
Moderated
No
Uptime
100.0%

Intelligence

21.7

62th percentile

Coding

Coding Index

Agentic

Agentic Index

Output speed

55 t/s

Median across providers

Latency

803ms

Time to first token

Cost per task

$0.09

Estimated

Benchmarks

Where the score comes from

The Intelligence Index is a composite. These are the underlying evaluations this model was actually measured on.

Evaluation scores

Percentage correct · higher is better

  • MMLU-Pro
    83.6%
  • GPQA Diamond
    75.1%
  • AIME 2025
    53.7%
  • LiveCodeBench
    52.9%
  • AA-LCR (long context)
    45.0%
  • IFBench
    41.2%
  • τ²-bench (Telecom)
    37.1%
  • SciCode
    32.1%
  • Terminal-Bench Hard
    31.8%
  • Humanity's Last Exam
    8.7%

An evaluation missing from this list was not run for this model — it is not a zero.

View as table
EvaluationScore
MMLU-Pro83.6%
GPQA Diamond75.1%
AIME 202553.7%
LiveCodeBench52.9%
AA-LCR (long context)45.0%
IFBench41.2%
τ²-bench (Telecom)37.1%
SciCode32.1%
Terminal-Bench Hard31.8%
Humanity's Last Exam8.7%

Against its peers

Intelligence Index · this model highlighted, nearest peers in grey

Peers are the models sitting closest on the Intelligence Index — the set you would realistically choose between.

View as table
ModelIntelligence
GLM 4.7 Flash23.3
Mercury 221.9
Qwen3.5-9B21.8
V3.1 Terminus21.7
Qwen3 Coder Next21.3
R120.4
North Mini Code (free)20.2
GPT-5 Nano20.1

Percentile among all indexed models

Intelligence62th
Mathematics50th
Terminal work75th

Pricing

What it costs to run

List prices per million tokens, plus what one representative task works out to.

List price

Input / 1M tokens
$0.27
Output / 1M tokens
$1
Cached input / 1M
$0.135
Blended 3:1
$0.453

One task, estimated

$0.09

Input tokens
50,000
Output tokens
80,000
Profile
Reasoning

Estimated from list pricing: 50K input tokens plus 80K output tokens for reasoning models (25K for non-reasoning).

Serving providers

Speed and latency figures are medians across these providers, so a widely-served model reports a blend rather than any single endpoint.