LFM2.5-24B-A2B vs LFM3-1.7B
LFM2.5-24B-A2B (LiquidAI) and LFM3-1.7B (LiquidAI) are listed at different prices. LFM2.5-24B-A2B costs less for input, $0.05 per 1M tokens against $0.08; LFM3-1.7B costs less for output, $0.20 per 1M tokens against $0.70. LFM3-1.7B has the larger context window, 262,144 tokens against 128,000 for LFM2.5-24B-A2B. Both accept text and produce text.
LiquidAI: LFM2.5-24B-A2B and LiquidAI: LFM3-1.7B both accept and produce text, support tools, structured outputs, response formats, and reasoning, and allow up to 8,192 output tokens. Their context limits are 128,000 and 262,144 tokens, respectively.
Compared from catalog data checked
Side by side
0.00 the better figure in a row| Specification | LFM2.5-24B-A2B | LFM3-1.7B |
|---|---|---|
| Maker | LiquidAI | LiquidAI |
| Added to catalog | 13 Oct 2026 | 6 Nov 2026 |
| Input, per 1M tokens | $0.05 | $0.08 |
| Output, per 1M tokens | $0.70 | $0.20 |
| Cached input read, per 1M | $0.025 | $0.02 |
| Context windowTokens | 128,000 | 262,144 |
| Max outputTokens | 8,192 | 8,192 |
| Accepts | text | text |
| Produces | text | text |
| Tool calling | Yes | Yes |
| Structured outputs | Yes | Yes |
| Reasoning effort control | No | No |
List prices through OpenRouter, in US dollars. Each model's page has its variants, price history and sample outputs.
What the same task costs
List price for one request, from the token counts shown. The same four tasks are priced on every model page.
| Task | LFM2.5-24B-A2B | LFM3-1.7B |
|---|---|---|
| Answer a short question500 in / 300 out | $0.0002 | $0.0001 |
| Summarise a 50-page PDF25,000 in / 800 out | $0.0018 | $0.0022 |
| Review a 2,000-line source file30,000 in / 2,000 out | $0.0029 | $0.0028 |
| Read a 300-page contract set150,000 in / 1,500 out | $0.0086 | $0.012 |
Which to choose
Choose LFM3-1.7B if
- Output tokens cost less: $0.20 against $0.70 per 1M
- Cached input costs less: $0.02 against $0.025 per 1M
- You need the larger context window: 262,144 tokens against 128,000
Neither maker's quality claims are repeated here. The reliable test is your own task: ask both the same question and read the two answers.