// head_to_head

GPT-5.4 vs MiMo-V2.6-Pro-UltraSpeed

Published LiveBench scores across all seven categories, live list pricing, context windows, and the measured cost of a point of capability — for both models, side by side.

OpenRouter + LiveBenchAll comparisonsFull leaderboard

GPT-5.4 and MiMo-V2.6-Pro-UltraSpeed are priced within ~10% of each other.

MiMo-V2.6-Pro-UltraSpeed does not have a published LiveBench run, so this comparison covers price, context and declared capabilities only. A missing score means "not evaluated", not "worse" — the right way to separate these two is an eval on your own workload.

openai

GPT-5.4

Blended / 1M
$5.63
Context
1.1M
Released
Mar 5, 2026
Overall score
78.0
reasoningtool callingimage inputfile inputprompt caching

xiaomi

MiMo-V2.6-Pro-UltraSpeed

Blended / 1M
$5.44
Context
1.0M
Released
Sep 21, 2026
Overall score
Not evaluated
reasoningtool callingimage inputvideo inputaudio inputprompt caching

Specs and pricing

MetricGPT-5.4MiMo-V2.6-Pro-UltraSpeed
LiveBench overall

Mean of the seven LiveBench category scores, 0–100. Higher is better.

78.0
Cost per point

Measured benchmark spend divided by overall score — dollars per point of capability.

$0.2198
Blended price / 1M

3:1 input:output mix, the usual shape of production traffic.

$5.63$5.44
Input price / 1M$2.50win$4.35
Output price / 1M$15.00$8.70win
Cached input / 1M

Price of an input token served from the prompt cache, where the provider publishes one.

$0.250$0.036win
Context window1.1M1.0M
Max output tokens128K131K

Benchmarks by category

An overall score averages away the thing you probably care about. These are the seven LiveBench categories scored separately — GPT-5.4 on top, MiMo-V2.6-Pro-UltraSpeed below, both out of 100.

Agentic coding
53.8
Coding
77.5
Reasoning
88.1
Mathematics
94.1
Data analysis
79.3
Language
82.6
Instruction following
70.2

What each one costs to run

Per-token prices are hard to feel. These are monthly list costs for both models across five workload shapes, using each provider's published cached-input rate where there is one.

WorkloadGPT-5.4MiMo-V2.6-Pro-UltraSpeed
Support chatbot

1.2K in / 400 out × 200K requests

$1,638/mo$1,429/mo
RAG assistant

8K in / 600 out × 100K requests

$2,000/mo$2,276/mo
Coding agent

40K in / 4K out × 20K requests

$1,940/mo$1,760/mo
Document extraction

20K in / 1.5K out × 50K requests

$3,513/mo$4,787/mo
Bulk classification

500 in / 20 out × 5M requests

$6,625/mo$9,588/mo
Run these two through the cost calculator

GPT-5.4 vs MiMo-V2.6-Pro-UltraSpeed FAQ

Which is better, GPT-5.4 or MiMo-V2.6-Pro-UltraSpeed?

GPT-5.4 and MiMo-V2.6-Pro-UltraSpeed are priced within ~10% of each other. MiMo-V2.6-Pro-UltraSpeed does not have a published LiveBench run, so this comparison covers price, context and declared capabilities only. A missing score means "not evaluated", not "worse" — the right way to separate these two is an eval on your own workload.

Is GPT-5.4 cheaper than MiMo-V2.6-Pro-UltraSpeed?

They cost about the same. Both land near $5.63 per million tokens on a 3:1 input:output blend, so price is unlikely to be the deciding factor between them.

Does GPT-5.4 or MiMo-V2.6-Pro-UltraSpeed have a bigger context window?

They are effectively the same — 1.1M for GPT-5.4 and 1.0M for MiMo-V2.6-Pro-UltraSpeed.

Do GPT-5.4 and MiMo-V2.6-Pro-UltraSpeed support prompt caching?

Both publish a cached-input rate: $0.250 per million for GPT-5.4 and $0.036 for MiMo-V2.6-Pro-UltraSpeed, against full input rates of $2.50 and $4.35. On a workload with a long stable prefix — a system prompt, a tool schema, a retrieved corpus — that changes the economics more than the headline price does.

Related comparisons

How these numbers are produced

  • Price — provider list price from OpenRouter, refreshed every 15 minutes. “Blended” is a 3:1 input:output mix.
  • ScoresLiveBench release 2026-06-25, using their own category map. Each model shows its strongest published run. A blank means “not evaluated”, never “bad”.
  • Cost per point — the measured dollars LiveBench spent on the run, divided by the score it earned.
  • “Win” — awarded only past a threshold: one full point on a benchmark score, 10% on a price, 25% on a context window. Anything tighter reports as a tie, because effort settings alone move a LiveBench score by more than that.

Published benchmarks rank models on someone else's tasks. Before committing, see LLM & agent evaluation for building an eval on your own.