Apple Foundation Models 3 vs Step 5 Preview
Apple · US | StepFun · China · Updated June 2026
Quick verdict
Pick Apple Foundation Models 3 for runs entirely on-device on iphone - no api call, no per-token cost, works offline or the 20b sparse 'core advanced' variant lifted on-device output quality from 23% to 46% user preference in apple's own testing versus the prior generation. Pick Step 5 Preview for sparse moe architecture — 600b total parameters, only 27b active per token — built for long-horizon agentic work across large codebases and documents or 1m-token context window (64k max output) at $1/$2.70 per million tokens, among the cheapest frontier-scale context windows available. On a tight budget at scale, Apple Foundation Models 3 is the value pick.
Apple Foundation Models 3 (Apple, US) and Step 5 Preview (StepFun, China) line up two different AI ecosystems against each other — a comparison that is as much about cost philosophy and openness as raw capability. Apple Foundation Models 3 is apple's on-device model family - genuinely small context and no API access, but private, offline, and free to run on every iPhone. Step 5 Preview is stepFun's September 20, 2026 flagship preview — a 600B-parameter sparse MoE model (27B active) with a 1M-token context window at $1/$2.70 per million tokens, open weights promised for mid-October. They diverge most on price and context window — each quantified below from the models' real specs.
Key differences at a glance
- ▸Context window: Step 5 Preview holds 244× more — 1M tokens (~1,500 pages) vs 4K (~6 pages). But effective recall usually fades long before the advertised ceiling, so the bigger number only helps if the model reasons over it.
- ▸Recency: Step 5 Preview is the newer model by about 4 months (released September 20, 2026), usually meaning fresher training data and capabilities.
- ▸Ecosystem: this is a US-vs-China matchup — they differ in pricing philosophy, data-residency options, and tooling ecosystems, not only benchmarks.
Side-by-side specs
| Spec | Apple Foundation Models 3 | Step 5 Preview |
|---|---|---|
| Provider | Apple (US) | StepFun (China) |
| Released | June 2026 | September 20, 2026 |
| Context window | 4K (~6 pages) | 1M tokens (~1,500 pages) |
| Price (in/out) | Not published | $1/$2.7 per 1M tokens |
| Open weight? | No — API only | No — API only |
| Modalities | text, image | text, image, code |
| SWE-Bench Verified | Not published | Not published |
| MRCR v2 @ 1M | Not published | Not published |
Who wins what
Runs entirely on-device on iPhone - no API call, no per-token cost, works offline
Apple Foundation Models 3
Apple Foundation Models 3 lists runs entirely on-device on iPhone - no API call, no per-token cost, works offline among its strengths; Step 5 Preview does not.
The 20B sparse 'Core Advanced' variant lifted on-device output quality from 23% to 46% user preference in Apple's own testing versus the prior generation
Apple Foundation Models 3
Apple Foundation Models 3 lists the 20B sparse 'Core Advanced' variant lifted on-device output quality from 23% to 46% user preference in Apple's own testing versus the prior generation among its strengths; Step 5 Preview does not.
Deep integration with iOS apps via the Foundation Models framework, now open to third-party LLM providers too
Apple Foundation Models 3
Step 5 Preview is comparatively weak here — trails closed frontier models on coding — 67.7% on DeepSWE v1.1 vs GPT-6 Astra's 74.1% and Claude Opus 5's 74.0%
Sparse MoE architecture — 600B total parameters, only 27B active per token — built for long-horizon agentic work across large codebases and documents
Step 5 Preview
Its 1M tokens window holds about 244× more than Apple Foundation Models 3's 4K in a single prompt.
1M-token context window (64K max output) at $1/$2.70 per million tokens, among the cheapest frontier-scale context windows available
Step 5 Preview
Apple Foundation Models 3 is comparatively weak here — a genuinely small context window (about 4,096 tokens, shared between input and output) - far below cloud frontier models
67.7% on DeepSWE v1.1, ahead of open-weight rivals Kimi K3 (67.5%) and GLM-5.3 (66.9%)
Step 5 Preview
StepFun's September 20, 2026 flagship preview — a 600B-parameter sparse MoE model (27B active) with a 1M-token context window at $1/$2.70 per million tokens, open weights promised for mid-October — and it carries the larger 1M tokens context.
Lowest cost at scale
Apple Foundation Models 3
Its weights are open, so at volume you pay for your own hardware instead of Step 5 Preview's $1/$2.7 per 1M tokens.
Largest single-prompt input
Step 5 Preview
Its 1M tokens window is about 244× larger than Apple Foundation Models 3's 4K, fitting roughly 1,500 pages in one prompt.
Which should you pick?
A cost-sensitive startup shipping high volume
→ Apple Foundation Models 3
At Not published it undercuts Step 5 Preview, and on millions of tokens that margin decides the monthly bill.
Someone analysing very long documents or codebases
→ Step 5 Preview
Larger 1M tokens window fits more in one prompt.
Anyone whose priority is runs entirely on-device on iphone - no api call, no per-token cost, works offline
→ Apple Foundation Models 3
It is specifically built for that.
Anyone whose priority is sparse moe architecture — 600b total parameters, only 27b active per token — built for long-horizon agentic work across large codebases and documents
→ Step 5 Preview
That is its strongest area.
An enterprise with regional data-residency rules
→ Apple Foundation Models 3 or Step 5 Preview
Origin (US vs China) affects where data is processed and which compliance regime applies — check the provider's terms for your region.
Apple Foundation Models 3: where it fits
Apple's on-device model family - genuinely small context and no API access, but private, offline, and free to run on every iPhone. Released June 2026 by Apple, it is built for runs entirely on-device on iPhone - no API call, no per-token cost, works offline, the 20B sparse 'Core Advanced' variant lifted on-device output quality from 23% to 46% user preference in Apple's own testing versus the prior generation, deep integration with iOS apps via the Foundation Models framework, now open to third-party LLM providers too, and zero marginal cost and strong privacy - nothing leaves the device.
Its trade-offs are real: a genuinely small context window (about 4,096 tokens, shared between input and output) - far below cloud frontier models, not comparable in raw capability to frontier cloud models - built for efficiency and privacy, not benchmark leadership, and no public API pricing since it isn't sold per-token - only usable within Apple's own platforms.
Step 5 Preview: where it fits
StepFun's September 20, 2026 flagship preview — a 600B-parameter sparse MoE model (27B active) with a 1M-token context window at $1/$2.70 per million tokens, open weights promised for mid-October. Released September 20, 2026 by StepFun, it is built for sparse MoE architecture — 600B total parameters, only 27B active per token — built for long-horizon agentic work across large codebases and documents, 1M-token context window (64K max output) at $1/$2.70 per million tokens, among the cheapest frontier-scale context windows available, 67.7% on DeepSWE v1.1, ahead of open-weight rivals Kimi K3 (67.5%) and GLM-5.3 (66.9%), and scores 44 on the Artificial Analysis Intelligence Index, ranking 27th of 653 models tracked.
Its trade-offs: trails closed frontier models on coding — 67.7% on DeepSWE v1.1 vs GPT-6 Astra's 74.1% and Claude Opus 5's 74.0%, open weights were promised for October 15, 2026 but hadn't shipped as of this writing — API-only for now despite the open-weight roadmap, a 'Preview' release, not StepFun's finished flagship — positioning and pricing may change once weights ship, and stepCodeBench and DeepSWE figures are StepFun's own reported numbers, not yet independently reproduced at scale. At $1 in / $2.7 out per million tokens, it sits in the budget price band.
The bottom line for this matchup
This is less "which is smarter" and more "which ecosystem fits." Apple Foundation Models 3 (US) and Step 5 Preview (China) differ on pricing philosophy, data-residency, and tooling as much as on raw scores. Apple Foundation Models 3 is the cheaper option, which matters at volume. The pragmatic move is to run one real task through both and judge the outputs against your own constraints — including where your data is allowed to be processed.
Want both Apple Foundation Models 3 and Step 5 Preview without two subscriptions? LumiChats gives you these plus 40+ models under one ₹69/day pass (about $1/day) — draft with one, cross-check with the other.
See pricingFrequently asked questions
Is Apple Foundation Models 3 or Step 5 Preview better for coding?
Public SWE-Bench figures are not available for either model, so the honest test is your own repository — run an identical real bug through both. By design, Apple Foundation Models 3 leans toward runs entirely on-device on iphone - no api call, no per-token cost, works offline while Step 5 Preview leans toward sparse moe architecture — 600b total parameters, only 27b active per token — built for long-horizon agentic work across large codebases and documents, and that positioning usually predicts which feels better on your codebase.
Which is cheaper, Apple Foundation Models 3 or Step 5 Preview?
Apple Foundation Models 3 is cheaper — Not published vs $1/$2.7 per 1M tokens.
Which has the bigger context window?
Step 5 Preview — 1M tokens vs 4K, about 244× larger. Useful only if the model actually reasons over the full window, which not all do.
Can I use both Apple Foundation Models 3 and Step 5 Preview together?
Yes — a multi-model platform like LumiChats gives you Apple Foundation Models 3, Step 5 Preview and 40+ others under one ₹69/day pass (about $1/day), so you can draft with one and cross-check with the other instead of buying two subscriptions.
Which is newer, Apple Foundation Models 3 or Step 5 Preview?
Step 5 Preview — released September 20, 2026, about 4 months after Apple Foundation Models 3.
Related comparisons
Specifications and benchmarks reflect publicly reported figures as of June 2026 and may change as providers release updates. Always verify on your own workload.