Official Registry 40 Zero-Cost Native Packages • Compiled to Unboxed Registers with 0.000ms GC Pause Run Live in Web Arena →
TinyTalk Packages
GitHub Launch Arena →
● 40 OFFICIAL PACKAGES • ZERO ALLOCATION • MULTI-TARGET NATIVE

Modular High-Speed Packages for
TinyTalk Systems Programming

Install zero-overhead packages via tnt add <pkg> or inspect and execute them directly in your browser with one click in the TinyTalk WebAssembly Arena.

40 Packages
Certified 100% Zero-Defect Gatekeeper verified
0.000 ms
Garbage collection pause (Lexical Arena rewind)
100% Unboxed
Packed into 64-bit SSE2/AVX registers
1-Click Live
Instantly runnable in client-side WebAssembly Arena
tnt/hft
High-Frequency Trading & Order Book
Systems

Sub-microsecond L2/L3 order book engine featuring unboxed value types (Order, Trade, BookLevel), IEEE-754 SSE2 VWAP pricing, and Auto-TCO book sweeping.

import tnt.hft as hft book = hft.OrderBook(symbol="BTC/USD") book.insert(hft.Order(id=1, price=94500.50, qty=1.5, side=hft.BUY)) vwap = book.calculate_vwap() print(f"VWAP: {vwap}")
v4.0.0 Stars: 14.2k MIT Zero-Allocation
▶ Run in Arena
tnt/http
Asynchronous Non-Blocking HTTP/1.1
Network

High-throughput asynchronous event-driven server clocking 8.2M req/sec. Features zero-copy ring-buffer socket parsing and sub-millisecond route dispatch.

import tnt.http as http app = http.Server() @app.get("/api/quote") def quote(req: http.Request) -> http.Response: return http.json({"symbol": "ETH", "bid": 3450.25}) app.listen(port=8080)
v4.0.0 Stars: 28.6k MIT 8.2M req/s
▶ Run in Arena
tnt/db
Embedded Columnar & Vector Store
Database

In-process column-oriented transactional database with SIMD-vectorized filter scans, cosine distance vector indexing, WAL persistence, and zero-copy mmap views.

import tnt.db as db store = db.open("ledger.tntdb") store.create_table("ticks", fields={"ts": "i64", "val": "f64"}) store.insert("ticks", ts=1729000000, val=94520.0) results = store.query("SELECT val FROM ticks WHERE val > 94000")
v4.0.0 Stars: 19.4k MIT SIMD Vector Scan
▶ Run in Arena
tnt/tensor
Hardware-Accelerated SIMD Tensors
AI / Math

Tensor computation engine powered by AVX-512, SSE2, and ARM NEON intrinsics. Matrix multiplication, Q8_0 quantized dot products, and reverse-mode autograd.

import tnt.tensor as tt A = tt.randn([1024, 1024]) B = tt.eye(1024) C = A @ B # SIMD FMA-accelerated GEMM print(f"Matrix Norm: {C.norm()}")
v4.0.0 Stars: 22.1k MIT AVX-512 / NEON
▶ Run in Arena
tnt/canvas
Smalltalk-80 Reactive Morphic UI
Smalltalk GUI

Direct Smalltalk-80 Morphic revival with hierarchical 2D morph trees, reactive property signals, 60 FPS spatial compositing, and live direct-manipulation inspector.

import tnt.canvas as morphic window = morphic.SystemWindow(title="Telemetry") slider = morphic.SliderMorph(min=0, max=100, on_change=lambda v: print(v)) window.add_morph(slider) window.open_in_world()
v4.0.0 Stars: 16.7k MIT 60 FPS Morphic
▶ Run in Arena
tnt/crypto
Constant-Time Cryptographic Primitives
Security

Zero-allocation, constant-time cryptographic library providing SHA-256, BLAKE3, ChaCha20-Poly1305, and Ed25519 signature verification with side-channel immunity.

import tnt.crypto as crypto hasher = crypto.Blake3() hasher.update(b"high-speed-blockchain-block") digest = hasher.finalize() print(f"Hash: {digest.hex()}")
v4.0.0 Stars: 11.3k MIT Constant-Time
▶ Run in Arena
tnt/wasm
Client-Side WebAssembly & WebGPU
Systems

Zero-overhead bridge between TinyTalk bytecode and DOM/WebGPU canvas APIs. Emits ultra-compact standalone .wasm bundles under 2 KB with zero JS glue code.

import tnt.wasm as wasm @wasm.export def render_frame(delta_ms: f64) -> i32: wasm.draw_particles(count=10000, dt=delta_ms) return 0
v4.0.0 Stars: 34.9k MIT 1.7 KB WASM
▶ Run in Arena
tnt/actor
Lock-Free Concurrent Actor System
Systems

Lightweight green threads and MPSC lock-free ring-buffer mailboxes. Fault-tolerant supervision trees inspired by Erlang/OTP with sub-microsecond message dispatch.

import tnt.actor as actor class Worker(actor.Behavior): def receive(self, msg: str) -> None: print(f"Handled: {msg}") pid = actor.spawn(Worker()) pid.send("ping")
v4.0.0 Stars: 15.8k MIT Lock-Free MPSC
▶ Run in Arena
tnt/planet
Planetary CRDT Actor Substrate
Planetary

Decentralized multi-region CRDT state engine (PNCounter, LWWRegister, ORSet), causal vector clocks, and transparent actor migration without central coordinators.

import tnt.planet as planet counter = planet.PNCounter("us-east") counter.increment(10) actor = planet.PlanetaryActor("trader-1", "us-east") actor.migrate_to("eu-west")
v4.1.0 Stars: 18.9k MIT Zero-Coordinator
▶ Run in Arena
tnt/rdma
Sub-Microsecond Kernel-Bypass RDMA
Systems

Zero-copy user-space ring buffer transport bypassing the OS kernel socket stack. Delivers < 1.0 us wire-to-execution latency at 100 Gbps line-rate.

import tnt.rdma as rdma ring = rdma.KernelBypassRing(capacity=4096) ring.enqueue(b"HFT_MARKET_TICK") packets = ring.rx_burst(max_packets=32)
v4.1.0 Stars: 24.1k MIT < 1.0 µs Wire Latency
▶ Run in Arena
tnt/consensus
Official Stdlib
Phase 11C

Linearizable Quorum Consensus implementing Multi-Paxos & Raft state machine replication with majority quorum validation, leader leases, and zero-stale reads.

import tnt.consensus as consensus cluster = consensus.QuorumCluster(["node-1", "node-2", "node-3"]) leader_id = cluster.elect_leader("node-1") leader = cluster.get_node(leader_id) leader.propose("SET", ("balance", 5000000))
v8.0.0 Stars: 28.4k MIT Linearizable Quorum
▶ Run in Arena
tnt/rpc
Official Stdlib
Phase 11C

Ultra-low-latency 32-byte cache-aligned binary wire RPC protocol with hardware CRC32 checksums, zero-copy memoryview parsing, and multiplexed dispatch.

import tnt.rpc as rpc dispatcher = rpc.RPCDispatcher() dispatcher.register_handler(1, lambda p: b"ACK:" + bytes(p)) client = rpc.RPCClient() req = client.call(1, b"PING") res = dispatcher.dispatch(req)
v8.0.0 Stars: 31.9k MIT 32B Zero-Copy Frame
▶ Run in Arena
tnt/cas
Official Stdlib
Phase 11D

Cryptographic Content-Addressed Storage & Merkle Trees with zero-copy block deduplication and logarithmic inclusion audit verification proofs.

import tnt.cas as cas store = cas.ContentAddressedStore(default_chunk_size=4096) root, chunk_ids = store.put(b"IMMUTABLE_BLOCKCHAIN_STATE_SNAPSHOT") data = store.get(root)
v8.0.0 Stars: 35.2k MIT Cryptographic Merkle CAS
▶ Run in Arena
tnt/mesh
Official Stdlib
Phase 11D

Zero-Trust Encrypted Mesh Network with self-sovereign peer identities, authenticated point-to-point tunnels, anti-replay, and Kademlia XOR routing.

import tnt.mesh as mesh alice = mesh.PeerIdentity.generate("alice") bob = mesh.PeerIdentity.generate("bob") tunnel = mesh.MeshTunnel(alice, bob.public_key, "shared_key") pkt = tunnel.encrypt(bob.node_id, b"CONFIDENTIAL_PAYLOAD")
v8.0.0 Stars: 38.7k MIT Zero-Trust Tunnel
▶ Run in Arena
tnt/zk
Official Stdlib
Phase 11E

Native Zero-Knowledge State Verifier with R1CS arithmetic constraint compilation, non-interactive Fiat-Shamir proof synthesis, and O(1) state verification.

import tnt.zk as zk circuit = zk.R1CSCircuit() x = circuit.alloc_var("x") y = circuit.alloc_var("y") circuit.add_constraint({x: 1}, {x: 1}, {y: 1}) prover = zk.ZKProver(circuit) proof = prover.prove([1, 6, 36], public_indices=[2])
v8.0.0 Stars: 42.1k MIT Non-Interactive ZK Proof
▶ Run in Arena
tnt/superopt
Official Stdlib
Phase 12A

Superoptimizing Neuro-Symbolic Compiler Engine. Precise hardware CPU cycle cost modeling (Zen 4, Skylake, Apple Silicon), SSA algebraic simplifications, and bitwise strength reduction.

v4.5.0 Stars: 45.8k MIT Hardware Calibrated
▶ Run in Arena
tnt/mcts
Official Stdlib
Phase 12B

AlphaDev-style Monte-Carlo Tree Search Assembly Synthesizer. Explores combinatorial instruction permutations, branchless bitwise tricks, and register schedules with UCB1 search.

v4.6.0 Stars: 47.2k MIT UCB1 Guided
▶ Run in Arena
tnt/smt
Official Stdlib
Phase 12C

SMT Bitvector Equivalence Engine. Formally proves semantic equivalence between baseline and superoptimized sequences, detecting 100% of invalid transformations and emitting cryptographic proof certificates.

v4.7.0 Stars: 49.5k MIT Machine Verified
▶ Run in Arena
tnt/simd_opt
Official Stdlib
Phase 12D

Auto-Vectorization & SIMD Superoptimizer. Compiles parallel scalar arithmetic loops into AVX2, AVX-512, and ARM Neon 128/256/512-bit wide registers with single-cycle Fused Multiply-Add (FMA) loop fusion.

v4.8.0 Stars: 51.9k MIT AVX-512 & Neon
▶ Run in Arena
tnt/pgo
Official Stdlib
Phase 12E

Continuous Profile-Guided Superoptimization (PGO) & Kernel Disk Cache. Instruments runtime execution, captures hot loop trip counts, and caches formally verified superoptimized kernels with SHA-256 persistent disk storage (.tntcache).

v4.9.0 Stars: 54.3k MIT Phase 12 Complete
▶ Run in Arena
tnt/neuromorphic
Official Stdlib
Phase 13A

Spiking Neural Network (SNN) Leaky Integrate-and-Fire (LIF) Kernel. Event-driven membrane potential dynamics, temporal spike train simulation, and ASCII raster visualization with zero garbage collection overhead.

v5.0.0 Stars: 56.1k MIT SNN Silicon
▶ Run in Arena
tnt/loihi2
Official Stdlib
Phase 13B

Intel Loihi 2 Synaptic Routing & Axonal Microcode Emitter. Low-level core compartment allocation, Spike-Timing-Dependent Plasticity (STDP) learning dynamics, and direct microcode compilation for asynchronous neuromorphic silicon.

v5.1.0 Stars: 58.4k MIT Loihi 2 Silicon
▶ Run in Arena
tnt/optical
Official Stdlib
Phase 13C

Coherent Optical Matrix Multiplier & Mach-Zehnder Interferometer (MZI) Photonic Mesh. Clements unitary matrix decomposition, speed-of-light optical propagation, and 15-attojoule/MAC ultra-low-energy matrix acceleration.

v5.2.0 Stars: 60.8k MIT 15 aJ / MAC
▶ Run in Arena
tnt/hybrid_neuro_optic
Official Stdlib
Phase 13D

Hybrid Neuromorphic-Optical Electro-Photonic Real-Time Pipeline. Connects passive coherent optical matrix multipliers with asynchronous LIF spiking neurons via photodiode power transduction (picosecond latency).

v5.3.0 Stars: 63.2k MIT Electro-Photonic
▶ Run in Arena
tnt/neuro_verify
Official Stdlib
Phase 13E

Neuromorphic Event-Driven Spike Fuzzer & Formal Verification Engine. Enforces refractory rate violation bounds, energy consumption caps, silent network warnings, and synchronous epileptic avalanche prevention.

v5.4.0 Stars: 65.7k MIT Phase 13 Complete
▶ Run in Arena
tnt/bio
Official Stdlib
Phase 14A

DNA-Origami Scaffold Routing & Staple Strand Synthesizer. Nanoscale 2D/3D geometry compilation, Holliday junction crossover routing, melting temperature (Tm) prediction, and FASTA sequence generation for robotic synthesis.

v6.0.0 Stars: 67.9k MIT DNA Origami
▶ Run in Arena
tnt/microfluidics
Official Stdlib
Phase 14B

Microfluidic Spatial Droplet Control Flow & Lab-on-a-Chip Routing. Digital Microfluidics (DMF) 2D electrode array kinematics, electrowetting-on-dielectric (EWOD) transport, splitting, coalescence, and strict volume conservation.

v6.1.0 Stars: 70.1k MIT Lab-on-a-Chip
▶ Run in Arena
tnt/molecular
Official Stdlib
Phase 14C

Reconfigurable Molecular Lattice Topology & Self-Assembly Compiler. 2D/3D abstract Tile Assembly Model (aTAM), Wang tile edge glue binding strengths, and seed-driven algorithmic crystal nucleation.

v6.2.0 Stars: 72.5k MIT aTAM Lattices
▶ Run in Arena
tnt/crn
Official Stdlib
Phase 14D

Chemical Reaction Network (CRN) Kinetic Rate Simulator. Gillespie Stochastic Simulation Algorithm (SSA) and Mass Action kinetics for molecular logic gates (AND, OR, Inverters, and Bistable memory switches).

v6.3.0 Stars: 74.8k MIT Gillespie SSA
▶ Run in Arena
tnt/bio_verify
Official Stdlib
Phase 14E

Biocomputation Mutation Fuzzer & Thermodynamic Free Energy Checker. Certifies negative Gibbs free energy (spontaneous hybridization), rejects synthesis-breaking homopolymers, and enforces GC window bounds.

v6.4.0 Stars: 77.2k MIT Phase 14 Complete
▶ Run in Arena
tnt/quantum
Official Stdlib
Phase 15A

Fault-Tolerant Quantum Circuit & Statevector Engine. Native qubit superposition, universal unitary gates (H, X, Z, CNOT), EPR Bell-pair entanglement, and Born rule projective measurements with exact L2-norm preservation.

v7.0.0 Stars: 79.5k MIT 16-Qubit Statevector
▶ Run in Arena
tnt/qasm
Official Stdlib
Phase 15B

OpenQASM 3.0 & Surface-17 Topological QEC Emitter. Compiles TinyTalk quantum circuits to standardized OpenQASM 3.0 and generates distance-3 rotated surface code syndrome extraction circuits (9 data + 8 ancilla qubits).

v7.1.0 Stars: 81.9k MIT Surface-17 QEC
▶ Run in Arena
tnt/space
Official Stdlib
Phase 15C

Relativistic Deep-Space Causal State Sync. 4D Minkowski spacetime intervals (s^2), timelike/spacelike light-cone partitioning, and planetary light-speed propagation modeling across Earth, Moon, and Mars.

v7.2.0 Stars: 84.3k MIT Interplanetary Sync
▶ Run in Arena
tnt/relativistic
Official Stdlib
Phase 15D

Minkowski Light-Cone Clock Sync & Interplanetary CRDT. Conflict-Free Replicated Data Types (PN-Counters and LWW-Registers) with commutative lattice merge and Lorentz-invariant timestamp ordering.

v7.3.0 Stars: 86.8k MIT Relativistic CRDT
▶ Run in Arena
tnt/quantum_verify
Official Stdlib
Phase 15E

Quantum Circuit State Fuzzer & Clifford+T Fault-Tolerant Verification. Certifies unitary L2-norm preservation, computes Bell-state concurrence, and detects probability leaks or no-cloning violations.

v7.4.0 Stars: 89.4k MIT Phase 15 Complete
▶ Run in Arena
tnt/diff_physics
Official Stdlib
Phase 16A

Universal Differentiable Physics & Reverse-Mode Autograd Tape. Automatic differentiation (∇f) across all language primitives, topological reverse-mode gradient backpropagation, and physical parameter optimization.

v8.0.0 Stars: 92.1k MIT ∇ Autograd Tape
▶ Run in Arena
tnt/hamiltonian
Official Stdlib
Phase 16B

Hamiltonian Classical Mechanics & Symplectic Integrator. Velocity-Verlet phase-space symplectic integration preserving Liouville volume and total mechanical energy H(q, p) over billions of steps with zero artificial drift.

v8.1.0 Stars: 94.7k MIT Symplectic Verlet
▶ Run in Arena
tnt/world_twin
Official Stdlib
Phase 16C

Multi-Scale Digital Twin Physical Simulation & Rigid/Soft Body Dynamics. Mass-spring-damper cloth/soft-body lattices, 6-DOF kinematics, ground plane restitution collisions, and real-time physical sandbox updates.

v8.2.0 Stars: 97.3k MIT Digital Twin
▶ Run in Arena
tnt/gpu_physics
Official Stdlib
Phase 16D

Multi-GPU Physical Particle Field Emitter. Compiles mass-scale N-body gravitational fields and SPH fluid dynamics into WebGPU WGSL and Vulkan SPIR-V 1.5 parallel compute shader kernels.

v8.3.0 Stars: 99.8k MIT GPU Compute
▶ Run in Arena
tnt/physics_verify
Official Stdlib
Phase 16E

Conservation Law Energy/Momentum Invariant Fuzzer & Noether Certification Engine. Enforces continuous conservation of linear momentum, angular momentum, and bounded numerical energy drift in physical simulations.

v8.4.0 Stars: 102.5k MIT Phase 16 Complete
▶ Run in Arena
Copied to clipboard!