API cost decision in 10 seconds

Step 5 Preview vs LFM2.5-2.6B

Pick LFM2.5-2.6B for lower cost; pick Step 5 Preview only if the larger context window matters more.

Page updated: 2026-10-24T22:08:08Z Data confirmed: 2026-10-25T00:08:13Z Prices normalized to USD per 1M tokens Sample workload: 1M input + 500K output

Budget verdict

Pick LFM2.5-2.6B for lower cost; pick Step 5 Preview only if the larger context window matters more.

On the standard 1M input plus 500K output workload, LFM2.5-2.6B is estimated at $0.6 vs $2.35 for Step 5 Preview, saving $1.75 (74.5% lower).

Cost-first pickLFM2.5-2.6B

Context-first pickStep 5 Preview

Sample savings$1.7574.5%

10x traffic gap$17.5

Step 5 Preview has more context, but LFM2.5-2.6B saves $1.75 on the standard workload. At 10x that traffic, the same price gap is about $17.5. Use the calculator below to replace the sample workload with your own token volume.

Cost sensitivity

Workload Sensitivity

Same prices, different token mixes.

LFM2.5-2.6B stays cheaper across input-heavy, balanced, and output-heavy sample workloads.

Workload shape	Token mix	Better pick	Step 5 Preview	LFM2.5-2.6B
Input-heavy / RAG	5M input + 500K output	LFM2.5-2.6B	$6.35	$2
Balanced workload	1M input + 1M output	LFM2.5-2.6B	$3.7	$0.85
Output-heavy chatbot	1M input + 5M output	LFM2.5-2.6B	$14.5	$2.85

Cheaper input LFM2.5-2.6B $1 vs $0.35 / 1M

LFM2.5-2.6B is $0.65 cheaper per 1M input tokens (65% lower; 2.86x difference).

Cheaper output LFM2.5-2.6B $2.7 vs $0.5 / 1M

LFM2.5-2.6B is $2.2 cheaper per 1M output tokens (81.5% lower; 5.4x difference).

Larger context Step 5 Preview 1.05M vs 128K

Step 5 Preview has 920.58K more context (8.19x larger).

Sample workload LFM2.5-2.6B $2.35 vs $0.6

LFM2.5-2.6B is $1.75 cheaper on the standard workload (74.5% lower).

Estimate your workload cost

Your Workload Cost

Prices are normalized to USD per 1M tokens.

Monthly input tokens Monthly output tokens

Step 5 Preview Calculating… Estimated API cost

LFM2.5-2.6B Calculating… Estimated API cost

Cheaper for this workload Calculating… Difference: calculating…

This estimate uses normalized public API pricing per 1M tokens. It is a planning aid, not a billing quote. Verify provider pricing, limits, and terms before production use.

Quick Decision

Verdict

LFM2.5-2.6B has the lower input price; LFM2.5-2.6B has the lower output price; Step 5 Preview offers the larger context window. For the 1M input plus 500K output sample, LFM2.5-2.6B is cheaper for the standard workload.

For a 1M input token plus 500K output token workload, the estimated API cost is $2.35 for Step 5 Preview and $0.6 for LFM2.5-2.6B.

Best Fit

Choose Step 5 Preview when you care most about larger context window.

Choose LFM2.5-2.6B when you care most about lower input-token price, and lower output-token price.

Decision Notes

On the standard 1M input plus 500K output workload, LFM2.5-2.6B is estimated at $0.6 vs $2.35 for Step 5 Preview, saving $1.75 (74.5% lower).
LFM2.5-2.6B is $1.75 cheaper on the standard workload (74.5% lower).
LFM2.5-2.6B is $0.65 cheaper per 1M input tokens (65% lower; 2.86x difference).
LFM2.5-2.6B is $2.2 cheaper per 1M output tokens (81.5% lower; 5.4x difference).
Step 5 Preview has 920.58K more context (8.19x larger).

Head-to-Head Specs

Feature	Step 5 Preview (StepFun)	LFM2.5-2.6B (LiquidAI)
Input Price prompt tokens per 1M	$1	$0.35
Completion Price per 1M tokens	$2.7	$0.5
Sample Workload Cost 1M input + 500K output	$2.35	$0.6
Context Window	1.05M	128K
Release Date	2026-10-08	2026-08-11

Use-Case Decision Matrix

Use case	Better pick	Why
Budget-constrained production	LFM2.5-2.6B	On the standard 1M input plus 500K output workload, LFM2.5-2.6B is estimated at $0.6 vs $2.35 for Step 5 Preview, saving $1.75 (74.5% lower).
High-volume input processing	LFM2.5-2.6B	Lower prompt-token price matters most when prompts, retrieved passages, or documents dominate the bill.
Long responses and chatbots	LFM2.5-2.6B	Lower output-token price matters most when assistants generate many completion tokens.
RAG or long-document work	Step 5 Preview	A larger context window leaves more room for retrieved passages, conversation history, or source files.

Related Alternatives

Same-provider lower-cost swaps

Step 3.5 Flash can replace Step 5 Preview when lower sample workload cost matters most: $0.25.
Step 3.7 Flash can replace Step 5 Preview when lower sample workload cost matters most: $0.77.
LFM2.5-2.6B (free) can replace LFM2.5-2.6B when lower sample workload cost matters most: $0.
LFM2.5-1.2B-Thinking (free) can replace LFM2.5-2.6B when lower sample workload cost matters most: $0.

Larger context near this budget

Grok 4.20 Multi-Agent offers 2M context with $2.5 sample workload cost.
Grok 4.20 offers 2M context with $2.5 sample workload cost.
Llama 4 Scout offers 1.31M context with $0.25 sample workload cost.
MiMo-V2.6-Flash offers 1.05M context with $0.28 sample workload cost.

Popular competitors

Kimi K3.1 Pro · MoonshotAI · #1
DeepSeek V4.1 Flash 1017 · DeepSeek · #2
DeepSeek V4.1 Flash · DeepSeek · #3
GLM 5.3 Flash · Z.ai · #4

Cheaper alternatives

Review low-cost models sorted by a standard 1M input plus 500K output workload.

Open cheapest models

Larger context alternatives

Find models with larger context windows for RAG, long documents, and codebase review.

Open largest context models

Provider catalogs

Compare models within provider hubs before choosing a final API vendor.

Open provider hubs

StepFun catalog

Review all tracked StepFun models before deciding whether this matchup is the right shortlist.

Open StepFun models

LiquidAI catalog

Check other LiquidAI models with comparable pricing, context, or release timing.

Open LiquidAI models

Step 5 Preview

Step 5 Preview is StepFun's flagship model for agentic work, built on a sparse Mixture-of-Experts architecture (27B active / 600B total parameters). It performs strongly in software engineering and professional...

LFM2.5-2.6B

LFM2.5-2.6B is a compact reasoning model from Liquid AI. It is suited for agent workflows, data extraction, RAG, and long-context processing. Liquid advises against using it for agentic coding or...