Nimind models

Three models. One workspace.

Nano for speed. Nexus for balance. Apex for depth. Pick one in the composer — and switch the moment the work changes.

Routes
3
Setup
None
Switching
Any time
Choose a model
3 routes

Nimind Nano

Fast routing

$0.25 / $1.00

in / out · 1M

Nimind Nexus

Balanced · Mixture of agents

$0.35 / $1.40

in / out · 1M

Nimind Apex

Complex synthesis

$5.00 / $25.00

in / out · 1M

Managed by Nimind · no API keysCompare
The lineup

Pick the depth the task needs.

All three routes run on Nimind — no provider keys to manage. Usage is priced per million tokens, with input and output rates shown on every route.

Fastest

Nimind

Nano

Fast routing

Input / 1M

$0.25

Output / 1M

$1.00

The quick answer, without the wait.

Nano takes the shortest sensible route through a task. It is the model to reach for when the work is clear, the answer should be quick, and you want to keep your momentum.

Best for

  • Quick drafts, rewrites and replies
  • Summaries, lookups and small edits
  • Everyday tasks that should not become a project

Choose it when You need an answer now and the task does not need deep deliberation.

Start with Nano
Recommended

Nimind

Nexus

Balanced · Mixture of agents

Input / 1M

$0.35

Output / 1M

$1.40

A specialist for every step.

Nexus balances speed with depth. It runs a mixture of agents, handing each part of a task to the approach best suited to it — useful when the job has more than one move.

Best for

  • Multi-step tasks with a clear deliverable
  • Research that mixes reading, tools and writing
  • Work that needs both momentum and judgement

Choose it when The job has several steps and you want a balance of speed and depth.

Start with Nexus
Deepest

Nimind

Apex

Complex synthesis

Input / 1M

$5.00

Output / 1M

$25.00

For the work that cannot be rushed.

Apex is built for the hardest jobs: long research, complex planning, careful code, and synthesis across many sources. It takes the extra passes when the output has to hold up.

Best for

  • Deep research and cross-source analysis
  • Complex planning, architecture and code
  • High-stakes writing that needs a careful review

Choose it when Quality matters more than latency, and the task deserves the extra pass.

Start with Apex

Prices are USD per 1 million tokens.

Benchmarks · Head to head

How it compares.

The same evaluations, with the well-known models each route was measured against. Our route is drawn solid; comparisons are muted. Scroll a chart sideways on a small screen.

Nano

Fast routing

vs 2 models

GPQA and HMMT are shown without tools.

Nexus

Balanced · Mixture of agents

vs 3 models

Vision and documents

OCRBench is shown normalised to 100.

Reasoning and code

Apex

Complex synthesis

vs 5 models

HLE is shown with tools; a missing bar means the comparison model did not report that benchmark.

Scores are from the published reports. All charts use percentages on a 0–100 scale; higher is better.

Apex, under the hood

Built for the long jobs.

Apex is a mixture-of-experts model with a million-token context and native image understanding. The size is there for the tasks that need it — not as a default for every request.

2.8T total · 104B active
Architecture
Mixture of experts
Total parameters
2.8T
Active per token
104B
Context length
1M tokens
Modality
Text + image
Experts
896 · 16 active
Layers
93
Vision encoder
401M params
Attention
69 KDA + 24 Gated MLA
Vocabulary
160K
Activation
SiTU-GLU
Quantization
MXFP4 / MXFP8

Not sure where to start?

Start with Nexus. Switch whenever you like.

Nexus balances speed and depth, which makes it a good default for real work. When a task turns trivial, move down to Nano. When it turns serious, move up to Apex.