Llama 3.1 Nemotron Ultra 253B vs Aya Vision 32B vs GLM-4.5-Flash
Too close to call on our weighted score (GLM-4.5-Flash 31, Llama 3.1 Nemotron Ultra 253B 31, Aya Vision 32B 15). The right pick depends on what you value most.
NVIDIALlama 3.1 Nemotron Ultra 253B
31/100- ECI—
- PriceFree / Free
- Context128K
CohereAya Vision 32B
15/100- ECI—
- Price—
- Context16K
Z.ai (Zhipu)GLM-4.5-Flash
31/100- ECI—
- PriceFree / Free
- Context131K
Too close to call
It is close. Our weighted score puts them within a point (GLM-4.5-Flash 31/100, Llama 3.1 Nemotron Ultra 253B 31/100, Aya Vision 32B 15/100), so choose by what matters most for your work: Llama 3.1 Nemotron Ultra 253B on price and GLM-4.5-Flash for long inputs. The score weighs inputs & features 60%, context window 40%. None of the shared benchmarks cover every model here yet, so capability is left out of this verdict.
- CapabilityNot enough dataNo independent benchmark covers every model here yet
- Lowest priceLlama 3.1 Nemotron Ultra 253B and GLM-4.5-FlashLlama 3.1 Nemotron Ultra 253B Free · GLM-4.5-Flash Free per 1M tokens (3:1 blend) · Aya Vision 32B unpriced
- Longest contextGLM-4.5-FlashGLM-4.5-Flash 131,072 · Llama 3.1 Nemotron Ultra 253B 128,000 · Aya Vision 32B 16,000 tokens
- Widest inputsAya Vision 32BLlama 3.1 Nemotron Ultra 253B: Text · Aya Vision 32B: Text, Images · GLM-4.5-Flash: Text
- Self-hostingLlama 3.1 Nemotron Ultra 253B and Aya Vision 32BPublishes downloadable weights (CC-BY-NC-4.0)
| Measure | Weight | Llama 3.1 Nemotron Ultra 253B | Aya Vision 32B | GLM-4.5-Flash |
|---|---|---|---|---|
| Inputs & features | 60% | 35 | 25 | 35 |
| Context window | 40% | 24 | 0 | 24 |
| Overall | 100% | 31/100 | 15/100 | 31/100 |
Left out because at least one model lacks the data: capability and price. The remaining weights were rescaled.
Every spec in one table
Highlighted cells lead their row. Dashes mean the data is not published.
| Specification | Llama 3.1 Nemotron Ultra 253BNVIDIA | Aya Vision 32BCohere | GLM-4.5-FlashZ.ai (Zhipu) |
|---|---|---|---|
| Capability | |||
| Capabilities Index (ECI) | — | — | — |
| ECI rank | — | — | — |
| Price per million tokens | |||
| Input | Free | — | Free |
| Output | Free | — | Free |
| Cached input | — | — | — |
| Blended (3:1) | Free | — | Free |
| Long-context rate | Same rate | — | Same rate |
| Price source | Official Nvidia API | — | Official Z.AI API |
| Limits | |||
| Context window | 128,000 tokens | 16,000 tokens | 131,072 tokens (best) |
| Max output | 8,192 tokens | 4,000 tokens | 98,304 tokens (best) |
| Inputs and features | |||
| Text | Yes | Yes | Yes |
| Images | No | Yes | No |
| PDFs | No | No | No |
| Audio | No | No | No |
| Video | No | No | No |
| Reasoning | Yes | No | Yes |
| Tool calling | Yes | No | Yes |
| Structured output | No | No | No |
| Availability | |||
| Weights | Open | OpenCC-BY-NC-4.0 | Proprietary |
| API model ID | nvidia/llama-3.1-nemotron-ultra-253b-v1 | c4ai-aya-vision-32b | glm-4.5-flash |
| API providers | 1 | 1 | 4 (best) |
| Released | Apr 7, 2025 | Mar 4, 2025 | Jul 28, 2025 |
| Knowledge cutoff | — | — | Apr 2025 |
What would a month cost?
Enter your expected volume in millions of tokens. List prices only; caching and batch discounts would lower these.
- Llama 3.1 Nemotron Ultra 253BFree
- Aya Vision 32B—
- GLM-4.5-FlashFree
Which should you choose?
Which is better: Llama 3.1 Nemotron Ultra 253B, Aya Vision 32B or GLM-4.5-Flash?
It is close. Our weighted score puts them within a point (GLM-4.5-Flash 31/100, Llama 3.1 Nemotron Ultra 253B 31/100, Aya Vision 32B 15/100), so choose by what matters most for your work: Llama 3.1 Nemotron Ultra 253B on price and GLM-4.5-Flash for long inputs. The score weighs inputs & features 60%, context window 40%. None of the shared benchmarks cover every model here yet, so capability is left out of this verdict.
Which is cheaper, Llama 3.1 Nemotron Ultra 253B, Aya Vision 32B or GLM-4.5-Flash?
Llama 3.1 Nemotron Ultra 253B is cheaper at Free input / Free output per million tokens (official Nvidia API price). GLM-4.5-Flash costs Free input / Free output per million tokens (official Z.AI API price). Llama 3.1 Nemotron Ultra 253B is listed as free. Aya Vision 32B has no published per-token price.
Which scores higher on benchmarks?
There is no independent benchmark that covers all three models yet. Llama 3.1 Nemotron Ultra 253B has not been scored yet, Aya Vision 32B has not been scored yet and GLM-4.5-Flash has not been scored yet.
Which is better for coding?
There are no published SWE-bench Verified results for Llama 3.1 Nemotron Ultra 253B, Aya Vision 32B and GLM-4.5-Flash yet, so there is no like-for-like coding score. Test both on a sample of your own tasks. Note that Aya Vision 32B does not support tool calling, which most coding agents need.
Which has the bigger context window?
GLM-4.5-Flash has the largest context window at 131,072 tokens, against 128,000 for Llama 3.1 Nemotron Ultra 253B and 16,000 for Aya Vision 32B. Maximum output per response: Llama 3.1 Nemotron Ultra 253B up to 8,192, Aya Vision 32B up to 4,000, GLM-4.5-Flash up to 98,304 tokens.
Which can read images, PDFs, audio or video?
Llama 3.1 Nemotron Ultra 253B accepts text; Aya Vision 32B accepts text and images; GLM-4.5-Flash accepts text. Aya Vision 32B handles the widest range of inputs.
Are any of these open source?
Llama 3.1 Nemotron Ultra 253B and Aya Vision 32B publishes its weights (CC-BY-NC-4.0) and can be self-hosted; GLM-4.5-Flash is proprietary.
Which is newer?
GLM-4.5-Flash is the newest, released Jul 28, 2025. Llama 3.1 Nemotron Ultra 253B came out Apr 7, 2025; Aya Vision 32B came out Mar 4, 2025. Knowledge cutoff: GLM-4.5-Flash Apr 2025.
How do you decide the winner?
Each model gets a 0–100 score on capability (50%, independent benchmark results); price (25%, blended price per million tokens (3 input : 1 output), log scale); inputs & features (15%, image, PDF, audio and video input, tool calling, structured output and reasoning); context window (10%, maximum tokens per request, log scale). Dimensions missing for any model are dropped and the remaining weights rescaled, so every model is judged on the same evidence. Specs and prices come from public model listings and the labs’ own API pages; capability scores come from independent benchmark runs. Data updated Oct 4, 2026.
