Graphite
SN43Solves complex graph optimisation problems using a decentralised network of solvers
Graphite is one of the more unusual Bittensor subnets, it starts from graph optimization problems like the traveling salesman problem, then layers in an organic portfolio workflow that makes the repo feel broader than a pure benchmark subnet.
// Graph optimization with an organic twist.
Graphite is a subnet built around solving graph optimization problems. The official README frames it around the traveling salesman problem and related routing-style tasks, while the miner and validator docs also expose an organic portfolio path tied to Taotrader and a yield server.
The simple version: It's like turning hard route-planning and portfolio-style optimization tasks into a competitive network.
Centralized equivalent: Think operations research software or optimization APIs, with a decentralized miner network competing on solution quality.
How it works:
- Miners run solver code that responds to validator requests and returns solutions to graph problems.
- Validators can run synthetic validators that generate graph problems, and the docs also describe organic validators tied to a backend and portfolio request flow.
- The problem it solves: Good optimization is useful anywhere routes, edges, constraints, or portfolio decisions matter. These problems get expensive fast as complexity grows.
- The opportunity: Graph optimization is a real category with applications in logistics, planning, and finance. If Graphite can turn that into a usable service layer, it has more substance than a subnet built only for leaderboard theater.
- The Bittensor advantage: Decentralized competition is a decent fit for optimization. Different miners can bring different heuristics and solvers, while validators keep pushing fresh synthetic or organic tasks into the network.
- Traction signals: Graphite has a live repo, a public website, and active mainnet docs. Market data is modest but stable, with price around 0.00617 TAO, market cap near 29,633 TAO, and slightly positive 7 day net flow of 33 TAO.
Category: Other (Graph Optimization) | Centralized Competitor: operations research solvers, route optimization APIs
Graphite exists in an unusual corner of Bittensor. The core README and website are consistent about the main story: decentralized graph optimization, especially TSP-style problems. The docs and protocol files then reveal a second layer, an organic portfolio and yield workflow, which suggests the subnet is trying to extend optimization beyond synthetic graph benchmarks into portfolio-style rebalancing tasks.
Mechanism:
The public repo supports two validator modes. Synthetic validators generate random graph problems using sampled graph sizes and structures, then send those challenges to miners. Organic validators are documented separately and connect to a backend flow that references proxy generation, databases, and a yield server. The protocol files also define portfolio request and yield data synapses, which supports the reading that Graphite is not only a TSP subnet anymore, even if graph optimization remains the main public framing.
That dual-track design is the interesting part. A pure synthetic subnet is easier to benchmark but harder to commercialize. A more organic optimization flow could eventually make Graphite more useful, but it also makes the product surface messier. Right now the safest reading is that Graphite is still fundamentally a graph optimization subnet with an added portfolio and yield layer under active development.
On market structure, Graphite looks healthier than many tiny subnets. Root proportion is about 0.158, which is low enough to suggest price is not being held up mainly by protocol subsidy. Liquidity and daily trading activity are both real, and 7 day net flow is still positive. The weaker side is concentration: gini is about 0.884, which says ownership or stake is concentrated.
If you are comparing it to cleaner product subnets, Graphite's challenge is clarity. The team has a real technical domain and an actual repo, but the public docs mix graph optimization, synthetic benchmarking, and Taotrader-linked organic flows. That does not kill the subnet, but it does mean the story is not as clean as it could be.
- Documentation drift. The README and website clearly emphasize graph optimization, but miner and validator docs also introduce Taotrader and yield-server portfolio flows. That makes the current product surface harder to parse.
- Concentration. Gini is elevated at about 0.884, which points to concentrated ownership or stake distribution.
- Execution risk. The repo warns that breaking changes may happen as the subnet continues active development. That is normal for an early subnet, but it raises operator friction.