Ember-1 vs Qwen2.5 VL 72B Instruct
Qwen2.5 VL 72B Instruct costs 15.0x less per output token than Ember-1 ($1.00 vs $15.00 per million). Updated October 22, 2026; no benchmark has tested both models yet.
- Coding
- 52.6
- Overall
- No data
- Output $/M
- $15.00
- Coding
- No data
- Overall
- No data
- Output $/M
- $1.00
Ember-1 or Qwen2.5 VL 72B Instruct: which to pick
Only Ember-1 has a coding score, 52.6 on the Vibe Coding Index; Qwen2.5 VL 72B Instruct has none, so the two cannot be ranked on coding.
Pick Qwen2.5 VL 72B Instruct for output tokens 15.0x cheaper ($1.00 vs $15.00 per million).
No benchmark has tested both yet. Ember-1 has results on Terminal-Bench 4 (Vals AI), Vibe Code Bench (Vals AI) and Code migration (Vals AI); Qwen2.5 VL 72B Instruct on OSWorld, Kagi LLM Benchmark and GeoBench. Compare them on price and context instead.
Which one for your work
Each card is decided by one number both models carry. Where only one has it, the card says so and makes no call.
Coding
Vibe Coding Index, six public coding benchmarks
No call
Only Ember-1 has a coding score, so no call is made.
One side not measured
Long documents
Listed context window
Ember-1
Ember-1 takes 1.05M tokens against 128K.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
Qwen2.5 VL 72B Instruct
Ember-1 $0.011, Qwen2.5 VL 72B Instruct $0.0013 per turn.
List prices
At a glance
- Coding index
- Ember-1 52.6 from 2 benchmarks, Qwen2.5 VL 72B Instruct not benchmarked.
- Widest category
- None yet: no category has been tested on both models.
- Widest single gap
- None yet: no benchmark has tested both models.
- Output price
- Ember-1 $15.00 against Qwen2.5 VL 72B Instruct $1.00 per million output tokens, 15.0x apart.
- Context window
- Ember-1 takes 1.05M tokens, Qwen2.5 VL 72B Instruct 128K.
Benchmark by benchmark
Every board either model has a result on, a category to a fold; bars are each result on a 0 to 100 scale. Only one model is measured on each, so no row names a leader.
CodingDirectional onlySix of these boards feed the Vibe Coding Index5 boards measured
| Benchmark | Ember-1 | Qwen2.5 VL 72B Instruct | Leader |
|---|---|---|---|
| Terminal-Bench 4 (Vals AI) | 20.7% | Not measured | Only Ember-1 measured |
| Vibe Code Bench (Vals AI) | 83.9% | Not measured | Only Ember-1 measured |
| Code migration (Vals AI) | 32.5% | Not measured | Only Ember-1 measured |
| IOI (Vals AI) | 46.3% | Not measured | Only Ember-1 measured |
| CyberBench (Vals AI) | 69.5% | Not measured | Only Ember-1 measured |
| Neither model has a result on the other 34 boards here | |||
Agentic workDirectional onlyPublic boards, shown as published, kept out of the index8 boards measured
| Benchmark | Ember-1 | Qwen2.5 VL 72B Instruct | Leader |
|---|---|---|---|
| Finance Agent (Vals AI) | 51.8% | Not measured | Only Ember-1 measured |
| Tax Agent Bench (Vals AI) | 31.1% | Not measured | Only Ember-1 measured |
| Legal Research Bench (Vals AI) | 44.7% | Not measured | Only Ember-1 measured |
| Excel Modeling Benchmark (Vals AI) | 63.4% | Not measured | Only Ember-1 measured |
| BioMysteryBench (Vals AI) | 67.0% | Not measured | Only Ember-1 measured |
| Terminal-Bench-Science (Vals AI) | 5.7% | Not measured | Only Ember-1 measured |
| Vals Science Index | 31.8% | Not measured | Only Ember-1 measured |
| OSWorld | Not measured | 5.0% | Only Qwen2.5 VL 72B Instruct measured |
| Neither model has a result on the other 33 boards here | |||
ReasoningDirectional onlyPublic boards, shown as published, kept out of the index2 boards measured
| Benchmark | Ember-1 | Qwen2.5 VL 72B Instruct | Leader |
|---|---|---|---|
| MysteryMechanism (Vals AI) | 22.5% | Not measured | Only Ember-1 measured |
| Kagi LLM Benchmark | Not measured | 36.0% | Only Qwen2.5 VL 72B Instruct measured |
| Neither model has a result on the other 20 boards here | |||
KnowledgeDirectional onlyPublic boards, shown as published, kept out of the index1 board measured
| Benchmark | Ember-1 | Qwen2.5 VL 72B Instruct | Leader |
|---|---|---|---|
| GeoBench | Not measured | 62.0% | Only Qwen2.5 VL 72B Instruct measured |
| Neither model has a result on the other 17 boards here | |||
GeneralDirectional onlyPublic boards, shown as published, kept out of the index4 boards measured
| Benchmark | Ember-1 | Qwen2.5 VL 72B Instruct | Leader |
|---|---|---|---|
| Public Benefits Bench (Vals AI) | 66.2% | Not measured | Only Ember-1 measured |
| LMArena Vision | Not measured | 1107 | Only Qwen2.5 VL 72B Instruct measured |
| LMArena Vision: English | Not measured | 1115 | Only Qwen2.5 VL 72B Instruct measured |
| LMArena Vision: Chinese | Not measured | 1107 | Only Qwen2.5 VL 72B Instruct measured |
| Neither model has a result on the other 42 boards here | |||
What each workload costs
Three fixed token mixes at each model's listed price, so per-token rates become a bill.
Chat turn
1K input + 500 output tokens
Qwen2.5 VL 72B Instruct costs less
Repository review
50K input + 3K output tokens
Qwen2.5 VL 72B Instruct costs less
Agent loop
200K input + 10K output tokens
Qwen2.5 VL 72B Instruct costs less
Listed OpenRouter rates, without cache or batch discounts. A workload that does not fit a model's context window would need splitting.
Specification differences
The leading value in a row is marked; a row with a missing value names no leader. Our two rows are marked in red.
| Specification | Fireworks Ember-1 | Qwen2.5 VL 72B Instruct |
|---|---|---|
| Maker | Fireworks | Qwen |
| Weights | No public weights listed | Open weights |
| Released | September 24, 2026 | February 1, 2025 |
| Context window | 1.05M | 128K |
| Input $/Mtok | $3.00 | $0.80 |
| Output $/Mtok | $15.00 | $1.00 |
| Coding index | 52.6 | Not benchmarked |
| Independent measurements behind it | 2 | 0 |
| Cost per solved task | Not computable | Not computable |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is Ember-1 better than Qwen2.5 VL 72B Instruct for coding?
Only Ember-1 has a coding score, 52.6 on the Vibe Coding Index; Qwen2.5 VL 72B Instruct has none, so the two cannot be ranked on coding. Pick Qwen2.5 VL 72B Instruct for output tokens 15.0x cheaper ($1.00 vs $15.00 per million).
Which is cheaper, Ember-1 or Qwen2.5 VL 72B Instruct?
Ember-1 costs $3.00 per million input tokens and $15.00 per million output tokens; Qwen2.5 VL 72B Instruct costs $0.80 and $1.00. Cost per solved task is known for neither of them.
Which has the bigger context window, Ember-1 or Qwen2.5 VL 72B Instruct?
Ember-1, with 1.05M tokens against 128K.
Has any benchmark tested both Ember-1 and Qwen2.5 VL 72B Instruct?
No benchmark has tested both yet. Ember-1 has results on Terminal-Bench 4 (Vals AI), Vibe Code Bench (Vals AI) and Code migration (Vals AI); Qwen2.5 VL 72B Instruct on OSWorld, Kagi LLM Benchmark and GeoBench. Compare them on price and context instead. Data as of October 22, 2026.