Decision support

Step 5 Preview vs Pareto 26.10

Compare canonical objects directly. The useful dimensions change with the object type, so model and hardware comparisons stay separate.

Models Hardware

Step 5 Preview

StepFun

Sources:models.devOpenRouter

Pareto 26.10

Unbiased

Sources:OpenRouter

Benchmark results

Arena Text
Arena rating · Higher is better
- Step 5 Preview
  1448
  Arena #83CI 1442 to 1455
- Pareto 26.10
  Not reported
Runs and sources
Step 5 Preview
default144895% CI 1442 to 1455 · 12,362 votes
Pareto 26.10
Not reported
Source:Arena Text leaderboard, 2026-11-09
Arena WebDev
Arena rating · Higher is better
- Step 5 Preview
  1573
  Arena #46CI 1565 to 1581
- Pareto 26.10
  Not reported
Runs and sources
Step 5 Preview
high157395% CI 1565 to 1581 · 7,583 votes
Pareto 26.10
Not reported
Source:Arena WebDev leaderboard, 2026-11-10
Arena Vision
Arena rating · Higher is better
- Step 5 Preview
  1245
  Arena #60CI 1234 to 1256
- Pareto 26.10
  Not reported
Runs and sources
Step 5 Preview
default124595% CI 1234 to 1256 · 3,523 votes
Pareto 26.10
Not reported
Source:Arena Vision leaderboard, 2026-11-09
Agent Arena
IPS score · Higher is better
- Step 5 Preview
  -0.037
  Arena #40CI -0.051 to -0.023
- Pareto 26.10
  Not reported
Runs and sources
Step 5 Preview
default-0.03795% CI -0.051 to -0.023 · 2,908,794 observations
Pareto 26.10
Not reported
Source:Agent Arena leaderboard, 2026-11-11

Each model's best published result is shown; "Not reported" means no matched evaluation, not a score of zero. Arena and DeepSWE ranks are the leaderboards' own; other ranks are among catalogue models with a score. Results from different harnesses or settings are not directly comparable.

Arena ratings: Arena (lmarena-ai/leaderboard-dataset), licensed CC BY 4.0. Ratings come from pairwise human votes and are only comparable within one leaderboard; the interval is Arena's 95% confidence interval. Model names are matched to the Vertical Tabs catalog.

Model specifications

Release date

September 16, 2026

October 15, 2026

Context window

Native context

—

Max output tokens

64,000

131,072

Knowledge cutoff

—

Open weights

Closed

Unknown

Parameter count

604B

—

API pricing · USD per 1M tokens

Input

$1.00

$0.80

Output

$2.70

$3.20

Cached input

$0.10

$0.03

Cache write

—

Capabilities and modalities

Input modalities

text, image, video

text, image

Output modalities

text

Tool calling

Supported

Reasoning option

Supported

—

Supported options

frequency_penalty, include_reasoning, logprobs, max_tokens, min_p, presence_penalty, reasoning, reasoning_effort, repetition_penalty, response_format, seed, stop, structured_outputs, temperature, tool_choice, tools, top_k, top_logprobs, top_p

max_tokens, temperature, tool_choice, tools, top_p

Capabilities

—

Architecture and availability

Architecture

—

Model families

—

Tokenizer

Other

Instruction format

—

License

—

Providers

Provider endpoints

Ollama variants

Metadata source

ai-research

A dash means the field is not reported. Provider-specific performance and pricing may differ.

Benchmark results

Arena Text

Arena WebDev

Arena Vision

Agent Arena

Model specifications

API pricing · USD per 1M tokens

Capabilities and modalities

Architecture and availability