Laguna-S-2.1-DFlash

Engine
vLLM
Listed
26 Oct 2026

Use on OpenRouter · Use on Vercel AI Gateway · Release blog post

Serves
Chat, Reasoning
Parameters
1.4B
Context
1,048,576 tokens
Weights
2.2 GB
License
Not stated

Capabilities

Read from the model's capability tags. Requests are admitted per tag.

Chat chat
  • POST /v1/chat/completions
  • POST /v1/responses

Endpoint task mode generate

Inputs
Text
Reasoning effort reasoning_effort
none, low, medium, high
Values a request may send; the model family sets the vocabulary.
Structured output structured_output
not yet probed
Admitted until the capability probe observes the model.
Forced tool choice forced_tool_choice
not yet probed
Admitted until the capability probe observes the model.

Launch settings

Every engine serving the model starts with these.

Set by the owner. auto: the serving node derives it. default: the platform's value, left in place.

Engine
vLLM
Context window --max-model-len
1,048,576 tokens
Quantization --quantization
none default
Weights load as stored, bfloat16.
KV cache dtype --kv-cache-dtype
auto
No flag: the checkpoint's KV quantization, else the model dtype.
Scheduling policy --scheduling-policy
priority default
Interactive requests ahead of batch work; same-class traffic in arrival order.
Tool-call parser --tool-call-parser
auto
Resolves to poolside_v1 from the architecture.
Reasoning parser --reasoning-parser
auto
Resolves to poolside_v1 from the architecture.
Trust remote code --trust-remote-code
on default
Python shipped in the bundle may run at load (the default).

Engine tuning

Every setting auto: the serving node derives each one.

Set by the owner. auto: the serving node derives it.

Memory and batching

GPU memory utilization --gpu-memory-utilization
auto
Max sequences --max-num-seqs
auto
Max batched tokens --max-num-batched-tokens
auto
Long-prefill threshold --long-prefill-token-threshold
auto

Execution

Enforce eager --enforce-eager
auto
Prefix caching --enable-prefix-caching
auto
Async scheduling --no-async-scheduling
auto
Custom all-reduce --disable-custom-all-reduce
auto

Speculative decoding

Draft tokens checked by the model in one pass.

Speculative decoding
not configured
An endpoint's own setting, or the serving node's default for the model family, applies.
Drafts for
Laguna-XS-2.1
This model is the speculative draft of a listed model.

Architecture

Read from the model's files at upload.

Architecture
laguna
Parameters
1,447,071,744
Layers num_hidden_layers
6
Hidden size hidden_size
3,072
Attention heads num_attention_heads
72 (8 KV)
Head dim head_dim
128
Vocabulary vocab_size
100,352
Native dtype torch_dtype
bfloat16
Weights on disk
2.2 GB, safetensors

Call it

Create an endpoint for the model, then send it a request in the OpenAI shape.

Shell
# Create an endpoint for this model, then provision it. # --tier-id is the service tier to run on; pick one at /endpoints/new. erectl endpoints create \ --name "My endpoint" \ --slug my-endpoint \ --model-id F563C65A-6E5C-465B-8644-5CA03A7427C0 \ --tier-id <tier-id> # provision takes the endpoint UUID returned by create erectl endpoints provision <endpoint-id> # Call it. Requests are routed by /<project-id>/<endpoint-slug>. curl https://api.erebine.ai/proj_ABC123/my-endpoint/v1/chat/completions \ -H "Content-Type: application/json" \ -H "Authorization: Bearer $EREBINE_API_KEY" \ -d '{ "model": "Laguna-S-2.1-DFlash", "messages": [{"role": "user", "content": "Hello, world!"}], "stream": true }'

Model card

Images are left out; its HTML tables are set as tables.

Laguna-S-2.1-DFlash

DFlash speculator (draft model) for
Laguna S 2.1, bf16.

  • Architecture: DFlashLagunaForCausalLM (6 sliding-attention layers, block_size 16).
  • Shares token embedding + lm_head with the target; draft_vocab_size == vocab_size (no d2t/t2d).
  • Loads under vLLM (native laguna_dflash) and TRT-LLM (pytorch DFlash backend) as the draft model in a speculative config.

Usage

vLLM
vllm serve --model poolside/Laguna-S-2.1 \
    --speculative-config '{"model":"poolside/Laguna-S-2.1-DFlash","num_speculative_tokens":7,"method":"dflash"}' \
    ...
SGLang
python -m sglang.launch_server --model-path poolside/Laguna-S-2.1 \
  --speculative-algorithm DFLASH \
  --speculative-draft-model-path poolside/Laguna-S-2.1-DFlash \
  ...
llama.cpp (GGUF)

A llama.cpp conversion of this draft model (laguna-s-2.1-DFlash-BF16.gguf) is
published in
poolside/Laguna-S-2.1-GGUF
alongside the target GGUFs. It embeds the target tokenizer and the DFlash metadata
(dflash.decoder_arch = laguna, capture layers, block size), so it works directly
as the -md draft model:

git clone --branch laguna https://github.com/poolsideai/llama.cpp
cd llama.cpp && cmake -B build && cmake --build build -j

./build/bin/llama-server \
  -m laguna-s-2.1-Q4_K_M.gguf \
  -md laguna-s-2.1-DFlash-BF16.gguf \
  --spec-type draft-dflash --spec-draft-n-max 7 \
  -fa on --jinja --port 8000

--spec-draft-n-max is clamped to the trained block size (15 draft tokens + 1).

Note. Requires Poolside's llama.cpp fork, branch laguna. Upstream llama.cpp ships the generic DFlash framework but not the Laguna decoder contract this draft model needs, and upstream PR ggml-org/llama.cpp#25165 covers the target architecture only.