vs
Modal vs Boundless
Modal rents per-second GPUs for code and weights you bring. Boundless sells ready open and closed models per token, plus reserved capacity its engineers run.
By The Subconscious Team · Updated
Modal vs Boundless: key differences
Modal runs any Python workload on GPUs from T4 to B300, billed per second from $0.59 an hour for a T4 to $3.95 for an H100 at list, with $30 of free credits each month. It has no model catalog, so teams bring weights and serving code. Boundless serves pre-tuned open models per token from prepaid credits, such as DeepSeek V4.1 Flash at $0.20 in and $1.00 out per million tokens, and resells GPT and Claude. On its tailored tier, Boundless engineers choose the model, tune precision, batching and caching, and operate it on reserved capacity.
Modal is the pick for custom weights, fine-tunes and non-LLM jobs like embeddings and OCR where control over the stack matters. Boundless fits teams that want LLM inference with nothing to deploy or operate, and it offers no fine-tuning. Its free two-week evaluation runs on real traffic against your current provider.
What Modal and Boundless do
Modal
Modal is serverless compute with GPUs attached. A developer decorates a Python function with the hardware it needs, such as gpu="H100", and Modal builds the container, schedules it, autoscales it and scales it back to zero. Billing runs per second with no minimum increment, from $0.59 an hour for a T4 to $3.95 for an H100 at list. Containers can hold up to 8 GPUs across T4 through B300.
Example models: none hosted by default; teams deploy their own, such as open LLMs on vLLM or Whisper
Full Modal 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 Modal or Boundless?
Modal vs Boundless at a glance
| Attribute | Modal | Boundless |
|---|---|---|
| Model access | Bring your own weights | Open weights, plus closed GPT and Claude |
| Flagship models | None hosted | GLM-5.3, DeepSeek V4.1 Flash, Kimi K3 |
| Speed | ~1s container boot | 342 tok/s on DeepSeek V4.1 Flash 1017 (Vercel) |
| Price | Per second; H100 $3.95/hr list | $0.04–$2.30 in, $0.21–$11.40 out per 1M (open) |
| Customization | Run any training code | Serving-stack tuning, no fine-tuning |
| Deployment | Serverless GPU containers | Serverless API, reserved tailored capacity |
| Long context | Depends on the model you deploy | 1M context on most models |
Frequently asked questions
What is the difference between Modal and Boundless?
Modal rents per-second GPUs for code and weights you bring. Boundless sells ready open and closed models per token, plus reserved capacity its engineers run.
When should I choose Modal over Boundless?
Custom weights and fine-tunes; Bursty per-second GPU jobs; Non-LLM work like OCR and embeddings.
When should I choose Boundless over Modal?
LLM tokens with nothing to deploy; Engineers who operate the deployment; Open and closed models on one API.
Is Modal or Boundless cheaper?
Modal: Per second; H100 $3.95/hr list. 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, Modal or Boundless?
Modal: Depends on the model you deploy. 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.