vs
Cerebras vs Boundless
Cerebras has the fastest published speed on a two-model shared catalog. Boundless serves a wider open and closed lineup and tunes reserved capacity.
By The Subconscious Team · Updated
Cerebras vs Boundless: key differences
Cerebras lists GPT-OSS 120B near 3,000 tokens per second at $0.35 in and $0.75 out per million tokens with 131K context, and its shared catalog adds only Qwen 3.8 27B. More models live on dedicated endpoints and through partners including Vercel, and OpenAI previewed an Ultrafast GPT-5.6 Sol on Cerebras at up to 750 output tokens per second. Boundless also serves models on Vercel AI Gateway, where Vercel measured 342 tokens per second on its DeepSeek V4.1 Flash 1017. Its own API carries open models with 1M context on most, plus Claude and a resold GPT-5.6 Sol at $2 in and $10 out.
Cerebras is the pick when generation is the wait, as in voice and live code autocomplete. Boundless fits teams that need longer context or a broader model choice, and those that want engineers to tune reserved capacity against contracted throughput and latency targets. It publishes no benchmarks of its own, so the free two-week evaluation is the place to test it.
What Cerebras and Boundless do
Cerebras
Cerebras builds a single chip the size of a silicon wafer, and its inference cloud is the fastest public host on the models it serves. Its developer table lists GPT-OSS 120B near 3,000 tokens per second, about six times Groq on the same weights, at $0.35 in and $0.75 out per million. The public shared catalog is thin, just GPT-OSS 120B and Qwen 3.8 27B since Gemma 4 31B moved to dedicated-only in September 2026. More model families live on dedicated endpoints and through partners like OpenRouter, Hugging Face, Vercel and AWS Marketplace.
Example models: GPT-OSS 120B, Qwen 3.8 27B
Full Cerebras profileBoundless
Boundless launched in September 2026 as an inference partner for AI-native startups. It sells two things. The Model API serves pre-tuned models on shared capacity, billed per token from prepaid credits with a budget per API key. Tailored inference reserves capacity for one customer, and Boundless engineers pick the model on the customer’s own evals, tune precision, batching, caching, speculative decoding and parallelism, then run the deployment against uptime, throughput and latency targets written into the contract. A free two-week evaluation tests those targets on real traffic against the customer’s current provider first.
Example models: GLM-5.3, DeepSeek V4.1 Flash, Kimi K3
Full Boundless profileShould you choose Cerebras or Boundless?
Cerebras vs Boundless at a glance
| Attribute | Cerebras | Boundless |
|---|---|---|
| Model access | Open weights | Open weights, plus closed GPT and Claude |
| Flagship models | GPT-OSS 120B, Qwen 3.8 27B | GLM-5.3, DeepSeek V4.1 Flash, Kimi K3 |
| Speed | ~3,000 tok/s on GPT-OSS 120B | 342 tok/s on DeepSeek V4.1 Flash 1017 (Vercel) |
| Price | $0.35 in, $0.75 out (GPT-OSS 120B) | $0.04–$2.30 in, $0.21–$11.40 out per 1M (open) |
| Customization | Unknown | Serving-stack tuning, no fine-tuning |
| Deployment | Shared API, dedicated, partners | Serverless API, reserved tailored capacity |
| Long context | 131K on GPT-OSS 120B, 128K on Qwen 3.8 27B | 1M context on most models |
Frequently asked questions
What is the difference between Cerebras and Boundless?
Cerebras has the fastest published speed on a two-model shared catalog. Boundless serves a wider open and closed lineup and tunes reserved capacity.
When should I choose Cerebras over Boundless?
Top published tokens per second; Voice and live code autocomplete; OpenAI's Ultrafast GPT-5.6 Sol preview.
When should I choose Boundless over Cerebras?
1M context on open models; GPT-5.6 Sol and Claude on one API; Tuned reserved capacity.
Is Cerebras or Boundless cheaper?
Cerebras: $0.35 in, $0.75 out (GPT-OSS 120B). Boundless: $0.04–$2.30 in, $0.21–$11.40 out per 1M (open). The cheaper choice depends on the model and workload.
Which has more context, Cerebras or Boundless?
Cerebras: 131K on GPT-OSS 120B, 128K on Qwen 3.8 27B. Boundless: 1M context on most models.
Related comparisons
Running long-horizon agents?
If your agents run past 200K tokens, compare both against Subconscious. Our inference stack treats a long-horizon trace as the primary workload, so speed, cost, and accuracy hold up deep into the trace.