Step 5 Preview
stepfun/step-5-preview
600B sparse-MoE agentic coding with 1M context.
cmd --model stepfun/step-5-preview
Intelligence index
43.7
Output speed
71.1 tok/s
Input
$1 /M
Output
$2.70 /M
Cache read
$0.05 /M
Agent-loop cost
$0.34 /M in
Context window
1M tokens
Released
September 18, 2026
Modalities
→
vs. the lineup
Step 5 Preview beside its stablemates and nearest rivals. The ◆ marks the best value in each column across every row shown.
pin a rival:
| Model | Intelligence | Coding | Speed | Input $/M | Output $/M | Blended $/M | Context |
|---|---|---|---|---|---|---|---|
| Muse Spark 1.3 Contributor | 48.1◆ | 75.8◆ | 247.5◆ | $0.10◆ | $0.20◆ | $0.13◆ | 1.05M◆ |
| Grok 4.6 | 44.3◆ | 76.8◆ | 70.3◆ | $2◆ | $6◆ | $3◆ | 500K◆ |
| Step 5 Preview ◆ | 43.7◆ | —◆ | 71.1◆ | $1◆ | $2.70◆ | $1.43◆ | 1M◆ |
| Gemini 3.5 Flash Lite | 22.2◆ | 49.3◆ | 370◆ | $0.30◆ | $2.50◆ | $0.85◆ | 1M◆ |
| Step 3.7 Flash | 19.5◆ | 39.6◆ | 196.7◆ | $0.20◆ | $1.15◆ | $0.44◆ | 256K◆ |
| Step 3.5 Flash | 17◆ | —◆ | —◆ | $0.10◆ | $0.30◆ | $0.15◆ | 1M◆ |
| DeepSeek V4 Pro (latest)pinned | 36◆ | 68.8◆ | 78.6◆ | $0.66◆ | $1.98◆ | $0.99◆ | 1M◆ |
| DeepSeek V4 Flash (latest)pinned | 34◆ | 69.1◆ | —◆ | $0.15◆ | $0.60◆ | $0.26◆ | 1M◆ |
coding performance
The Intelligence Index and its sub-scores, ranked against every scored model in the catalog. A metric that has not been measured for Step 5 Preview has been left empty.
Intelligence Index
43.7
#15 of 67 scored
Long-context reasoning
88.3
reasoning across a long context
SciCode
58.9
scientific coding
usage calculator
How far a month of credits goes on Step 5 Preview.
Input tokensfresh prompt
800Output tokensmodel reply
180Cache read tokensre-read context
50Kcost / request $0.0038 · in $1 · out $2.70 · cache $0.05 per M
fresh input 21%output 13%cache reads 66%
Requests / 30 days
2.6K
$10 credits ÷ $0.0038 per request
~528 quick fixes~106 bug fixes~18 feature PRs
what real work costs
Real coding tasks priced end to end on Step 5 Preview, from a quick lookup to a full-repo agent run.
One agent task
$0.08
180K in at 75% cache hit, 12K out
What you pay
$0.44 /M
all-in across every token that task touched
Sticker input
$1 /M
cache reads bill at $0.05 /M instead
| Task | Tokens in · out | Step 5 Preview | Muse Spark 1.3 Contributor | Claude Haiku 4.5 |
|---|---|---|---|---|
| Quick lookup / one-liner | 8K · 1K | $0.0039 | $0.0003 | $0.0065 |
| Review a 500-line PR | 60K · 4K | $0.03 | $0.0027 | $0.04 |
| Fix a bug (agent loop) | 180K · 12K | $0.08 | $0.0072 | $0.12 |
| Refactor a module | 320K · 20K | $0.16 | $0.01 | $0.21 |
| Full-repo agent run | 900K · 45K | $0.35 | $0.03 | $0.49 |
frequently asked
How much does Step 5 Preview cost?
$1/M input and $2.70/M output, cache reads $0.05/M. In an agent loop most input is cache-read, so the effective input rate is about $0.34/M.Which plan do I need?
Available on Go and above.
How do I switch to it?
Run
cmd --model stepfun/step-5-preview, or type /model in a session and pick it. You can switch mid-session without losing context.Ship code that matches your taste
Command Code is the AI coding agent that continuously learns your taste. Start for $1.
Benchmarks from Artificial Analysis (v4.3)commandcode.ai/models/step-5-preview
