Providers & APIs
The external services Clawlas integrates — exchanges, LLMs, market & on-chain data, RPC, aggregators, orchestration and infrastructure — with roles, pricing, limits and official docs.
Clawlas is provider-independent: every external service is an adapter behind a stable internal interface. This is the working catalogue of those services — what each is for, what it costs, and where to verify the current terms.
Exchanges, backend & LLMs
| Provider | Role | Pricing / limits | Docs |
|---|---|---|---|
| CCXT / CCXT Pro | Central CEX/DEX integration + WebSockets | Free (WS merged into the package) | docs.ccxt.com |
| Convex (self-hosted) | Reactive app backend & operational state | No per-request bill; VPS + storage only | docs.convex.dev/self-hosting |
| Google Gemini | Cheap classification → mid reasoning | Flash-Lite $0.10 → Pro $1.25 in / 1M | ai.google.dev/gemini-api/docs/pricing |
| OpenAI | Structured tasks → advanced analysis | GPT-5 mini $0.25 → GPT-5.5 $5.00 in / 1M | developers.openai.com/api/docs/pricing |
| Anthropic Claude | Fast tool use → deep strategy review | Haiku 4.5 $1.00 → Sonnet 4.6 $3.00 in / 1M | docs.anthropic.com/en/docs/about-claude/pricing |
| DeepSeek | Cost-efficient reasoning (open-weight, self-hostable) | V3 $0.14 in · $0.28 out / 1M | api-docs.deepseek.com/quick_start/pricing |
| Alibaba Qwen | Cheap multilingual extraction → analysis (open-weight) | Flash $0.19 → Max $1.25 in / 1M | alibabacloud.com/help/en/model-studio/model-pricing |
| Moonshot (Kimi) | Agentic tool use & long-context synthesis (open-weight) | Kimi K2.5 $0.60 in · $3.00 out / 1M | platform.moonshot.ai/docs/pricing |
Prices are per 1M input tokens at the reference date and change frequently — recheck before committing. Output prices and cached-input discounts are on the cost-model page. The open-weight providers (DeepSeek, Qwen, Kimi) are dual-use: a cheap hosted API now, and self-hostable weights through the gateway below once volume justifies a GPU.
Model gateway & open-source serving
| Provider | Role | Pricing | Docs |
|---|---|---|---|
| LiteLLM | Unified gateway, fallbacks, budgets (policy stays in Clawlas) | Open source | docs.litellm.ai/docs/routing |
| Ollama | Local dev & lightweight quantized models | Open source | docs.ollama.com |
| llama.cpp | Efficient quantized inference (CPU/GPU) | Open source | github.com/ggml-org/llama.cpp |
| vLLM | High-throughput GPU serving (future dedicated GPU) | Open source | docs.vllm.ai |
Market, on-chain & social data
| Provider | Role | Pricing / limits | Docs |
|---|---|---|---|
| DEX Screener | Pool/pair discovery & liquidity | ~60 rpm profiles · ~300 rpm pairs/search | docs.dexscreener.com/api/reference |
| CoinGecko / GeckoTerminal | Metadata, reconciliation, on-chain pools | Free Demo API (quota) + MCP | docs.coingecko.com |
| Birdeye | Faster on-chain discovery, security, WS | Free → Lite $39 → Starter $99 → Premium $199 → Business $499 | docs.birdeye.so/docs/pricing |
| X API | Early social signal (curated accounts) | Pay-per-use credits; per-endpoint pricing | docs.x.com/x-api/getting-started/pricing |
| SearXNG (self-hosted) | Project research & fallback web search | No licence fee | docs.searxng.org |
Discovery starts on the free sources (DEX Screener, CoinGecko Demo, public RPC). Birdeye and X are added only when their data measurably improves opportunity capture, each behind a budget with auto-stop.
On-chain execution & RPC
| Provider | Role | Pricing / limits | Docs |
|---|---|---|---|
| Jupiter | Solana routing & execution | Free API | dev.jup.ag |
| Helius | Solana RPC & data | Free tier → paid; recheck | helius.dev/pricing |
| 0x | EVM swap aggregation (executable tx) | Per dashboard/API; log zeroExFee | docs.0x.org |
| 1inch | EVM swap aggregation (alternative) | Per portal pricing | portal.1inch.dev/pricing |
| Alchemy | Multi-chain EVM RPC | Free 30M CU/mo, 25 rps → pay-as-you-go | alchemy.com/pricing |
Aggregators (Jupiter, 0x, 1inch) are interchangeable behind the VenueAdapter interface; RPC providers run primary + fallback. Premium RPC is justified only by measured landing-rate or latency impact on net PnL.
Orchestration, infrastructure & ops
| Provider / tool | Role | Pricing / limits | Docs |
|---|---|---|---|
| LangGraph | Durable agent orchestration & human-in-the-loop | Open source | docs.langchain.com/oss/javascript/langgraph/overview |
| Valkey | Event streams, cache, locks, rate limits | Open source | valkey.io |
| BullMQ / Dramatiq / Celery | Job queues, retries, scheduling | Open source | docs.bullmq.io |
| Hostinger VPS (KVM 4) | Shared host (app, Convex, workers) | $0 incremental (already owned) | hostinger.com/vps-hosting |
| Docker / Compose | Service isolation & deployment | Free for this use | docs.docker.com |
| GitHub Actions | CI/CD | Free tier 2,000 min/mo | github.com/pricing |
| Cloudflare R2 | Off-site backups | 10 GB free, then $0.015/GB-mo, free egress | developers.cloudflare.com/r2/pricing |
| Restic | Encrypted backups → R2 | Open source | restic.net |
| Uptime Kuma | Endpoint & worker monitoring | Open source | github.com/louislam/uptime-kuma |
| Resend | Email alerts (Telegram is primary) | Per plan | resend.com/pricing |