Modal vs General Compute

Modal rents GPUs by the second for models you deploy. General Compute sells per-token open models and dedicated ASIC racks.

By The Subconscious Team · Updated October 27, 2026

Modal vs General Compute: 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 a month. It has no model catalog, so teams bring weights and serving code. General Compute serves four open models through an OpenAI-compatible API at $0.21 to $0.28 in per million tokens, with GPU prefill and ASIC decode. It reports over 1,000 tokens per second on GPT-OSS 120B in its own benchmark.

Modal is the pick for custom models, embeddings, OCR and batch jobs where control over the stack matters. General Compute fits teams that want fast tokens on its models with nothing to deploy. It also takes LoRA or fine-tuned checkpoints, so a custom LLM can run on ASIC decode without managing GPUs. For steady large loads it sells reserved racks, bare metal and tokens-per-second SLAs.

What Modal and General Compute 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 profile

General Compute

General Compute is a San Francisco inference cloud built on alternative chips. It splits each request across hardware, with NVIDIA GPUs handling prefill and ASICs from Cerebras, SambaNova, Positron AI and d-Matrix handling decode. In its own open benchmark on GPT-OSS 120B it measured 1,025 to 1,234 output tokens per second against 100 to 121 on an unnamed GPU cloud, and up to 16.2x faster end to end. Founders Finn Puklowski and Jason Goodison raised a $15M seed led by FUSE, an up to $400M debt facility from Upper90 and, on October 27, 2026, a $100M Series A co-led by Blackbird and Abstract.

Example models: MiniMax M2.7, DeepSeek V3.2, GPT-OSS 120B

Full General Compute profile

Should you choose Modal or General Compute?

Modal

Choose Modal for

Custom and non-LLM models
Bursty per-second GPU work
Batch and media jobs

General Compute

Choose General Compute for

Fast open-model tokens with nothing to deploy
Fine-tuned LLMs on ASIC decode
Reserved racks for steady load

Modal vs General Compute at a glance

Attribute	Modal	General Compute
Model access	Bring your own weights	Open weights
Flagship models	None hosted	MiniMax M2.7, DeepSeek V3.2, GPT-OSS 120B
Speed	~1s container boot	1,000+ tok/s on GPT-OSS 120B (vendor)
Price	Per second; H100 $3.95/hr list	$0.21–$0.28 in, $0.38–$1.20 out per 1M
Customization	Run any training code	Custom LoRA and fine-tuned checkpoints
Deployment	Serverless GPU containers	Serverless API, dedicated racks, bare metal
Long context	Depends on the model you deploy	192K (MiniMax M2.7)

Frequently asked questions

What is the difference between Modal and General Compute?

Modal rents GPUs by the second for models you deploy. General Compute sells per-token open models and dedicated ASIC racks.

When should I choose Modal over General Compute?

Custom and non-LLM models; Bursty per-second GPU work; Batch and media jobs.

When should I choose General Compute over Modal?

Fast open-model tokens with nothing to deploy; Fine-tuned LLMs on ASIC decode; Reserved racks for steady load.

Is Modal or General Compute cheaper?

Modal: Per second; H100 $3.95/hr list. General Compute: $0.21–$0.28 in, $0.38–$1.20 out per 1M. The cheaper choice depends on the model and workload.

Which has more context, Modal or General Compute?

Modal: Depends on the model you deploy. General Compute: 192K (MiniMax M2.7).

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.

Subconscious vs Modal Subconscious vs General Compute