sllm – Split a GPU node with other developers, unlimited tokens
Running DeepSeek V3 (685B) requires 8×H100 GPUs which is about $14k/month. Most developers only need 15-25 tok/s. sllm lets you join a cohort of developers sharing a dedicated node. You reserve a spot with your card, and nobody is charged until the cohort fills. Prices start at $5/mo for smaller models.The LLMs are completely private (we don't log any traffic).The API is OpenAI-compatible (we run vLLM), so you just swap the base URL. Currently offering a few models.
FL score
out of 100
Verdict
high confidence
Competition
12
competitors found, emerging market, funded players
Trend
No signal yet
A platform enabling individual developers to share GPU nodes for large LLM inference with unlimited tokens via a cohort model, tackling high costs and complexity.
The pain
The gap
Build angle
Strengths
- Addresses a severe, high-cost problem for a specific developer audience.
- The 'unlimited tokens' for a low fixed price is a very attractive value proposition.
- OpenAI-compatible API (vLLM) lowers adoption friction.
- The LLM inference market is experiencing explosive growth.
Risks
- Extremely high competition from well-funded GPU cloud providers and decentralized networks.
- Significant technical complexity and operational overhead for a solo builder to manage GPU infrastructure reliably.
- Uncertainty around cohort fill rates and the economic viability of 'unlimited tokens' pricing.
- Risk of 'noisy neighbor' problem still existing within cohorts, undermining the core value proposition.
- High capital expenditure (or rental costs) required for procuring/renting H100 GPUs.
Next steps
- Conduct in-depth interviews with target developers to confirm 'unlimited tokens' and the 'cohort' model are desired over existing hourly rates or managed APIs.
- Validate the exact technical requirements and potential solutions to truly mitigate the 'noisy neighbor' problem within a shared node.
- Launch a waitlist to gauge interest and assess the speed at which cohorts would realistically fill for specific LLMs.
- Model out the unit economics of the 'unlimited tokens' pricing to ensure profitability once cohorts are filled, considering peak usage patterns.
Fly Labs Method
Is the pain real, is there a gap, is it the right time, can one person build it.
- Problem clarity
- 27
- Solution gap
- 15
- Willingness to pay
- 14
- Buildability
- 9
Addresses a real and expensive problem for developers but enters a crowded market with high build complexity and strong incumbents, limiting a solo builder's competitive advantage without a hyper-specific niche.
Value Equation
Dream outcome and how likely it feels, against the time and effort it costs.
Strong market demand but faces significant competition and high operational complexity, making differentiation and pricing power difficult for a new entrant.
One-Person Business
Curiosity pull, identity fit, and a path from free value to paid for a solo creator.
Addresses a clear problem for a reachable audience but the high complexity and the need for significant specialized expertise make it unsuitable for a solo creator aiming for simplicity and rapid iteration.
Viral Frameworks
Hook strength, shareability, and how cheaply it can be tested.
A clear value proposition for a specific audience but significant business model and technical assumptions, coupled with challenging distribution in a crowded market, make it high-risk for a micro-SaaS.
Builder Lens
Evidence the problem exists, timing, defensibility, and a model that fits on a napkin.
Addresses a desperate need for affordable large LLM access for specific users, with clear market demand, but faces challenges in validating the narrowest wedge and competing in a crowded space.
Why this verdict
Five lenses, one composite. How scoring works
The angle
This weekend
- Conduct in-depth interviews with target developers to confirm 'unlimited tokens' and the 'cohort' model are desired over existing hourly rates or managed APIs.
- Validate the exact technical requirements and potential solutions to truly mitigate the 'noisy neighbor' problem within a shared node.
- Launch a waitlist to gauge interest and assess the speed at which cohorts would realistically fill for specific LLMs.
- Model out the unit economics of the 'unlimited tokens' pricing to ensure profitability once cohorts are filled, considering peak usage patterns.
Who is already there, emerging market
RunPod offers a developer-centric GPU cloud with both secure data center compute and a community cloud leveraging peer-to-peer resources.
Pricing: Community Cloud RTX 3090 from $0.22/hr; Secure Cloud RTX 4090 from $0.34/hr; H100 PCIe priced at $1.99/hr.
SaladCloud operates a large distributed network of consumer GPUs, pooling idle capacity for affordable large-scale compute.
Pricing: GTX 1050 Ti from $0.015/hr; RTX 4090 starting at $0.16/hr; H100 NVL at $0.99/hr. Enterprise pricing for high-volume use available upon contact.
- Vast.aiVisit
Vast.ai is a peer-to-peer GPU marketplace where individuals and data centers rent out unused capacity, with prices fluctuating based on supply and demand.
Pricing: RTX 3090 from $0.13/hr; RTX 4090 from $0.31/hr.
Akash Network uses a decentralized model to connect users directly with compute providers through a permissionless marketplace with an open bidding system.
Pricing: RTX 4090 around $0.40/hr; H100 GPUs from $1.20/hr.
- Jarvis LabsVisit
Jarvis Labs offers on-demand access to the latest NVIDIA GPUs like H100 and A100 at competitive hourly rates with no minimum commitments.
Pricing: H100 SXM $2.69/hour; A100-80GB $1.49/hour; RTX5000 $0.39/hour.
- FAR AI (FAR Labs)Visit
FAR AI is a distributed AI inference network connecting consumer and enterprise GPUs into a single network, routing requests to optimal nodes and allowing GPU owners to earn income.
Pricing: Pricing not yet publicly available; node registrations are open with developer API access scheduled for Q2.
Hosted.ai provides a GPU-as-a-service software stack that optimizes the management and consumption of GPU resources through pooling, multi-tenant workload placement, and overcommitment.
Pricing: Pricing not publicly available; focuses on improving utilization for service providers.
CoreWeave is a GPU-native cloud provider purpose-built for training and inference at scale, optimized for modern AI stacks.
Pricing: H100 x 8 priced at $1.99/GPU/hour with a 12-month commitment; on-demand H100 at $3.39/hour.
Lambda Labs offers a GPU cloud platform tailored for AI developers and researchers, providing on-demand access to high-end GPUs.
Pricing: Pricing details not explicitly provided, but noted for 'research-friendly pricing with pre-configured ML frameworks'.
Inferless is a serverless GPU platform enabling developers to deploy machine learning models with a focus on cost-effective, high-performance AI inference.
Pricing: Their 'shared GPU pricing is not something I have seen anywhere' according to a user, but specific numbers are not public.
- AgenticalVisit
Agentical is a platform enabling users to share and rent their GPU computing power via a browser using WebGPU for AI chat or other tasks.
Pricing: Pricing not explicitly stated, but focuses on a community-driven approach to computing.
- GPUStackVisit
GPUStack is an open-source GPU cluster manager for running LLMs, allowing organizations to create unified clusters from various GPUs.
Pricing: Open-source, no direct pricing as it's a self-hosted solution.
What they charge
Recent news
- Show HN: sllm – Split a GPU node with other developers, unlimited tokens | Hacker News
Hacker News, April 04 2026
- FAR Labs Opens FAR AI Node Registrations to Tap 3B Idle GPUs - Tech Times
Tech Times, April 02 2026
- Hosted.ai: $19 Million Raised By GPU Infrastructure Software Platform To Transform AI Economics - Pulse 2.0
Pulse 2.0, March 25 2026
- CoreWeave lands $8.5B from Wall Street banks to scale its AI cloud - Tech Funding News
Tech Funding News, April 01 2026
- AI cloud company Together AI, which rents out Nvidia chips, pursues $1B in fresh funding: report | Seeking Alpha
Seeking Alpha, March 05 2026
Market signals
The GPU for AI market is experiencing explosive growth, projected to reach $26.09 billion by 2032 from $3.34 billion in 2023. Significant funding rounds are common, with cloud GPU platforms raising $4.85 billion in equity funding across 12 rounds in 2025 (as of November). The market is driven by the increasing adoption of AI, machine learning, and data analytics across industries, with a strong emphasis on optimizing inference costs and GPU utilization.
What frustrates people
- Noisy neighbor problem: Degradation of Time To First Token (TTFT) when other users on a shared node spin up heavy 24/7 jobs, despite continuous batching techniques like vLLM.
- Resource contention: Physical limits with shared VRAM/compute leading to performance issues in shared GPU environments.
- Payment failures: Concern about what happens if some users' payment methods fail once a cohort fills and charges are initiated.
- High costs of dedicated GPUs: The significant investment required for high-end GPUs like 8x H100s ($14k/month) makes them inaccessible for many individual developers.
- Complexity of self-hosting: Organizations face challenges in installing and managing complex clustering software and AI tool stacks for private LLM deployments.
Semble – Code search for agents that uses 98% fewer tokens than grep
Hey HN! We (Stephan and Thomas) recently open-sourced Semble. We kept running into the same problem while using Claude Code on large codebases: when the agent can't find something directly, it falls back to grep, reading full files or launching subagents. This uses a lot of tokens, and often still misses the relevant code. There are existing tools for this, but they were either too slow to index on demand, needed API keys, or had poor retrieval quality.Semble is our solution for this. It combines static Model2Vec embeddings (using our latest static model: potion-code-16M) with BM25, fused via RRF and reranked with code-aware signals. Everything runs on CPU since there's no transformers involved. On our benchmark of ~1250 query/document pairs across 63 repos and 19 languages, it uses 98% fewer tokens than grep+read and reaches 99% of the retrieval quality of a 137M-parameter code-trained transformer, while being ~200x faster.Main features:- Token-efficient: 98% fewer tokens than grep+read- Fast: ~250ms to index a typical repo on our benchmark, ~1.5ms per query on CPU (very large repos may take longer)- Accurate: 0.854 NDCG@10, 99% of the best transformer setup we tested- MCP server: drop-in for Claude Code, Cursor, Codex, OpenCode- Zero config: no API keys, no GPU, no external servicesInstall in Claude Code with: claude mcp add semble -s user -- uvx --from "semble[mcp]" sembleOr check our README for other installation instructions, benchmarks, and methodology:Semble: https://github.com/MinishLab/sembleBenchmarks: https://github.com/MinishLab/semble/tree/main/benchmarksModel: https://huggingface.co/minishlab/potion-code-16MLet us know if you have any feedback or questions!
AI
Postgres extension for BM25 relevance-ranked full-text search
Last summer we faced a conundrum at my company, Tiger Data, a Postgres cloud vendor whose main business is in timeseries data. We were trying to grow our business towards emerging AI-centric workloads and wanted to provide a state-of-the-art hybrid search stack in Postgres. We'd already built pgvectorscale in house with the goal of scaling semantic search beyond pgvector's main memory limitations. We just needed a scalable ranked keyword search solution too.The problem: core Postgres doesn't provide this; the leading Postgres BM25 extension, ParadeDB, is guarded behind AGPL; developing our own extension appeared daunting. We'd need a small team of sharp engineers and 6-12 months, I figured. And we'd probably still fall short of the performance of a mature system like Parade/Tantivy.Or would we? I'd be experimenting long enough with AI-boosted development at that point to realize that with the latest tools (Claude Code + Opus) and an experienced hand (I've been working in database systems internals for 25 years now), the old time estimates pretty much go out the window.I told our CTO I thought I could solo the project in one quarter. This raised some eyebrows.It did take a little more time than that (two quarters), and we got some real help from the community (amazing!) after open-sourcing the pre-release. But I'm thrilled/exhausted today to share that pg_textsearch v1.0 is freely available via open source (Postgres license), on Tiger Data cloud, and hopefully soon, a hyperscalar near you:https://github.com/timescale/pg_textsearchIn the blog post accompanying the release, I overview the architecture and present benchmark results using MS-MARCO. To my surprise, we were not only able to meet Parade/Tantivy's query performance, but exceed it substantially, measuring a 4.7x advantage on query throughput at scale:https://www.tigerdata.com/blog/pg-textsearch-bm25-fu
AI
Consus (YC F24) – LLM gateway for companies handling sensitive data
Hey Hacker News! I'm Eric, the founder of Consus. We route all AI requests to ITAR restricted and controlled unclassified information (CUI) compliant infrastructure. Consus itself satisfies ITAR restrictions, and we recently completed our FedRAMP equivalency audit against the 410 FedRAMP High controls. FedRAMP High is the U.S. government’s strictest security baseline for cloud services handling sensitive unclassified data. A breach of this data could cause severe or catastrophic harm. Achieving FedRAMP equivalency, or authorization, is required before processing CUI information.Demo: https://youtu.be/6MYxJe6P4oYAerospace and defense is experiencing rapid growth. Unfortunately, most companies are struggling to use AI because of necessary data protections & compliance risk. These companies can move even faster if they could use the AI tools that most other industries take for granted.I spent most of my career prior to Consus in aerospace & defense. I launched national security payloads at SpaceX, and was an early employee at Castelion. I’ve watched first hand how compliance red-tape makes it difficult for companies to adopt new technologies and trust vendors that touch their sensitive data. For example, I worked with a satellite manufacturer that doesn’t allow any AI tools, even for non-sensitive work.Even as AI has become extremely useful over the past 9-12 months, I was surprised by how many aerospace & defense companies refused to touch it. They could see how it could be helpful, but they were scared about model providers training on their data, violating an ITAR rule, or risk losing new business because of a failed cyber assessment. These are all rational fears.There were a lot of ways to solve some/all of these problems, but we focused on building a compliant AI inference layer that can "plug" into any AI tool or workflow. Just as OpenRouter integrates with most tools, Consus can connect to tools like ChatGPT or Clau
AI
Questions about this idea?
FlyBot reads the scoring and gives you a second opinion on “sllm – Split a GPU node with other developers, unlimited tokens”.